Parenteral Quality Control

Marcel Dekker, Inc.
New York

Basel
Parenteral
Quality Control
Sterility, Pyrogen, Particulate,
and Package Integrity Testing
Third Edition, Revised and Expanded
Baxter Pharmaceutical Solutions LLC
Bloomington, Indiana, U.S.A
Michael J. Akers and Daniel S. Larrimore
RxPax, LLC
Bridgewater, New Jersey, U.S.A
Dana Morton Guazzo
Copyright 2002 by Marcel Dekker, Inc. All Rights Reserved.
Copyright 2003 Marcel Dekker, Inc.
Library of Congress Cataloging-in-Publication Data
Acatalog recordfor this bookis available fromthe Library of Congress.
ISBN: 0-8247-0885-7
The second edition was authored by Michael J. Akers (Marcel Dek-
ker, 1993).
This book is printed on acid-free paper.
Headquarters
Marcel Dekker, Inc.
270 Madison Avenue, New York, NY 10016
tel: 212-696-9000; fax: 212-685-4540
Eastern Hemisphere Distribution
Marcel Dekker AG
Hutgasse 4, Postfach 812, CH-4001 Basel, Switzerland
tel: 41-61-260-6300; fax: 41-61-260-6333
World Wide Web
http:/ /www.dekker.com
The publisher offers discounts on this book when ordered in bulk
quantities. For more information, write to Special Sales/Profes-
sional Marketing at the headquarters address above.
Copyright 2003 by Marcel Dekker, Inc. All Rights Re-
served.
Neither this book nor any part may be reproduced or transmitted
in any form or by any means, electronic or mechanical, including
photocopying, microlming, and recording, or by any information
storage and retrieval system, without permission in writing fromthe
publisher.
Current printing (last digit):
10 9 8 7 6 5 4 3 2 1
PRINTED IN THE UNITED STATES OF AMERICA
To my mother
D.L.
Preface
Drug products administered by injection are characterized by
three qualities possessed by no other type of pharmaceutical
dosage form: sterility, freedomfrompyrogenicity, and freedom
from particulate matter. The achievement of sterile, nonpyro-
genic, and particulate-free parenteral products provides a sig-
nicant challenge to the ingenuity and creativity of parenteral
scientists and technologists. Of equal challenge are the suc-
cessful application and performance of analytical testing pro-
cedures to verify the claims of parenteral products that they
are indeed, sterile, pyrogen-free, and free from visible particu-
late contamination.
Ofcial compendial tests for sterility, pyrogens, and par-
ticulates evoke widespread controversy regarding their reli-
ability, sensitivity, and applicability. While impressive tech-
nological advances have been made in the production of
parenteral products, the testing for the quality of these prod-
ucts involves relatively simple procedures. One of the objec-
tives of this book is to critique the adequacy of current meth-
ods for sterility, pyrogen, particulate, and leak testing and to
review future trends and improved technology in these areas.
v
vi Preface
The focus of this third edition of Parenteral Quality Con-
trol is to update the QC practitioner on the ever-changing re-
quirements of various regulatory bodies and to educate the
practitioner on newtechnologies associated with testing meth-
ods of the nished product form. Methodologies for performing
tests that meet U.S. and European requirements are empha-
sized.
The late 1990s brought many regulatory changes to the
pharmaceutical industry. Not only are inspectors from FDA
and European regulatory bodies focusing on quality in manu-
facturing areas but these regulators are also expecting higher
standards, especially with respect to sterility assurance, endo-
toxin, and particulate control in the quality control laboratory.
While the MCA, USP, EP, and FDA all have different expecta-
tions, the latter part of the 1990s brought great harmonization
where quality control is concerned with the previously men-
tioned areas of testing. These updates and harmonization ef-
fort are discussed in this book.
New to this edition is an expanded discussion of barrier
isolation technology, which has become the standard in steril-
ity testing. While no regulations exist for the use of barrier
isolation systems, industry expectations exist for qualication
of these systems when used for nished-product sterility test-
ing. Regulators are especially concerned with the generation
of a false-negative result when surface decontamination is em-
ployed with the use of a barrier isolation system. Methods of
validation of such a system are discussed. Finally, the chapter
on package integrity testing has been updated to educate the
reader as to developments in recent technology. Updated ref-
erences have been added throughout the chapters to provide
recent information.
I would like to thank KarimAbdelkader, Ryan Akers, and
the marketing department at Baxter Pharmaceutical Solu-
tions LLC for the cover photograph.
Daniel S. Larrimore
Contents
Preface v
1. Sterility Testing 1
Introduction 1
Sterility and Sterility Test Regulations 3
Sampling for Sterility Testing 8
Culture Media 14
Time and Temperature of Incubation 20
Sterility Test Methods 26
Sterility Retesting 31
FDA Guidelines on Sterility Testing 32
Sterility Testing of Different Sterile Products 40
Sterility Testing of Antibiotics and Proteins 51
Control in Sterility Testing 52
Validation of the Sterility Test 57
Limitations of the USP/NF Referee Sterility
Test 58
vii
viii Contents
Isolation Chambers and Robotic Sterility Test
Units 67
Support Techniques and Procedures for
Sterility Assurance 75
Alternatives to the Compendial Sterility Test 102
References 109
2. Pyrogen Testing 119
Introduction 119
History 122
Specic Requirements of the USP Rabbit
Pyrogen Test 124
The Limulus Amebocyte Lysate Test 140
Modications of Rabbit and LAL Tests for
Pyrogens 182
References 183
3. Particulate Matter Testing 197
Introduction 197
Background of Particulate Matter Concerns in
Parenteral Products 198
Nature and Source of Particulates 202
The Reality of Particulate Matter
Contamination in Parenteral Solutions 203
Particulate Matter Standards 206
Visual Inspection: Manual Methods 210
Visual Inspection: Automatic Methods 221
Current Issues with Visible Particle Detection
and Inspection Procedures 233
USP Test for Subvisible Particulate Matter in
Large-Volume Injections for Single-Dose
Infusion and Small-Volume Injections 235
Electronic Particle Counters 239
Light Obscuration Particle Count Test 248
Contents ix
Microscopic Particle Count Test 252
Automated Microscopic Particulate Matter Test 257
Comparison of Microscopic and Electronic
Particle-Counting Methods 260
Current Issues with Electronic Particle
Counters 266
Factors Affecting Accurate Particle Testing 269
International Compendia Standards for
Particulate Matter Content in Parenteral
Solutions 271
References 273
4. Package Integrity Testing 281
Introduction 281
Regulations and Guidances 283
Leakage 287
Establishing Leak Rate Specications 294
Package Integrity Test Methods 307
The Changing Pharmaceutical Industry 349
Package Integrity and the Product Life Cycle 354
Conclusion 361
References 363
Appendix I Example of Standard Operating
Procedure for Sterility Testing by
Direct Inoculation 369
Appendix II Example of Standard Operating
Procedure for Sterility Testing by
Membrane Filtration 373
Appendix III Aseptic Procedures at the Laminar
Flow Workbench 379
Appendix I
Example of Standard
Operating Procedure for
Sterility Testing by Direct
Inoculation*
Purpose: To describe the USP test procedure for sterility test-
ing by direct inoculation.
Equipment and Supplies
1. Trypticase soy broth (TSB) mediumsterile test tubes of
appropriate size, one for each sample plus three controls
2. Fluid thioglycollate medium (FTM)sterile test tubes of
appropriate size, one for each sample plus three controls
3. Sterile syringes or pipets, one for each sample
4. One incubator at 3035C
5. One incubator at 2025C
* Reprinted by permission of Dr. K. E. Avia, College of Pharmacy, University of Ten-
nessee, Memphis, Tennessee.
369
370 Appendix I
6. Laminar ow (LF) workbench
7. Sterile Tyvek gown
8. Sterile disposable cap
9. Sterile mask
10. 70% Alcohol
11. Betadine or pHisohex
12. Urethane wipes
Procedure
Steps Comments
1. The operator shall wear a
sterile coat, mask, hood, and
disposable cap. The operator
shall wash hands thoroughly
with a disinfectant scrub be-
fore donning sterile gloves.
2. Wipe inside, top, and counter Take care to avoid splashing
surfaces of LF workbench surface of HEPA lter. Sur-
with suitable surface disinfec- face disinfectant is usually
tant. 70% alcohol, but others may
be used and should be from
time to time.
3. Wipe all exposed surfaces of
vials, ampules, tubes of cul-
ture media, and other con-
tainers with surface disinfec-
tant before placing them in
LF workbench.
4. All sterile items having an Alternatively, the outer wrap-
outer wrapping should have ping may be wiped with a
the wrapping removed at the disinfectant prior to placing
edge of the LF workbench in the workbench, but this is
and the sterile inner item in- less desirable.
troduced aseptically into the
workbench.
Appendix I 371
Steps Comments
5. After all supply items have The usual hand disinfectant
been introduced into the LF is pHisohex or betadine. Ster-
workbench, the operator ile latex or PVC gloves may
should change to a new pair be worn, but are not re-
of sterile gloves, or prefera- quired.
bly, partner will then per-
form the critical aseptic
steps using uncontaminated
gloves.
6. For vials, remove protective Leave damp, but there
seal and disinfect exposed rub- should be no pool of disinfec-
ber closure with alcohol wipe. tant.
7. For ampules, break neck at Avoid splattering of HEPA
score mark, pointing toward lter with liquids. Do not
side of workbench rather place hands between ltered
than HEPA lter. air source and critical site.
8. Use a sterile syringe or pipet Refer to USP sterility test
to transfer the appropriate procedures for appropriate
volume of product to each volumes of media and prod-
test tube containing either uct inocula.
sterile TSB or FTM.
9. After all required product These tubes will serve as pos-
samples have been inocu- itive controls to show that
lated, inoculate one addi- the test organism grows in
tional tube of TSB and FTM the presence of the product.
with product. Then, inocu-
late each tube with a loopful
of standard test organism cul-
ture of a spore strip.
10. Two additional tubes of TSB The inoculated tubes will
and FTM should be used as show that the culture media
controls. One tube of each support growth of microor-
medium should be inoculated ganisms and the blank tubes
with a loopful of test organ- will conrm the sterility of
ism and the other tube left the culture media.
uninoculated.
372 Appendix I
Steps Comments
11. Incubate the samples at ap- Incubate FTM at 3035C
propriate temperatures and and TSB at 2025C (room
observe after 3, 7, and 10 temperature [RT]). Longer in-
days for presence of micro- cubation may be required at
bial growth. times to permit slow growers
to develop.
12. Record results on quality con-
trol record sheet.
13. Sterilize used culture media
and clean tubes after incuba-
tion.
Appendix II
Example of Standard
Operating Procedure for
Sterility Testing by
Membrane Filtration*
Purpose: To describe a method for the use of membrane lters
in the sterility testing of sterile products.
Equipment and Supplies
1. Sterility test lter holder unitMillipore Steril, Falcon
unit
2. Membrane lter, 0.45 mm, 47 mm, hydrophobic edge
3. Sterile needles, syringes, or administration sets20
4. Sterile trypticase soy broth (TSB), 100-ml tubes3
5. Sterile uid thioglycollate medium (FTM), 100-ml
tubes3
* Reprinted by permission of Dr. K. E. Avis, College of Pharmacy, University of Ten-
373
374 Appendix II
6. Sterile diluting uid, 100 ml3
7. Sterile scissors
8. Sterile forceps, smooth tip, 46 inch stainless steel
9. Sterile disposable gloves
10. Alcohol, 70% denatured
11. Urethane wipes
12. Laminar ow hood (LFH)
13. Blue plastic base with hole for Falcon unit
14. Test samples
15. Sterile gown, cap, and mask
Procedure
Steps Comments
1. The operator shall wear a
sterile coat, mask, hood, and
disposable cap. The operator
shall wash hands thoroughly
with a disinfectant scrub be-
fore donning sterile gloves.
2. Wipe inside, top, and counter Take care to avoid splashing
surfaces of LF workbench surface of HEPA lter. Sur-
with suitable surface disinfec- face disinfectant is usually
tant. 70% alcohol, but others may
be used from time to time.
3. Wipe all exposed surfaces of
vials, ampules, tubes of cul-
ture media, and other con-
tainers with surface disinfec-
tant before placing them in
LF workbench.
4. All sterile items having an Alternatively, the outer wrap-
outer wrapping should have ping may be wiped with a
the wrapping removed at the disinfectant prior to placing
edge of the LF workbench in the workbench, but this is
and the sterile inner item in- less desirable.
troduced aseptically into the
workbench.
Steps Comments
5. After all supply items have The usual hand disinfectant
been introduced into LF is pHisohex or betadine. Ster-
workbench, the operator ile latex or PVC gloves may
should change to a new pair be worn, but are not re-
of sterile gloves, or prefera- quired.
bly, partner will then per-
form the critical aseptic steps
using uncontaminated gloves.
6. Remove overseals from the Do not leave excess alcohol
necks of test samples, previ- on closure. Allow the clo-
ously disinfected ampules, sures to air dry in the LF
multidose vials or large-vol- hood.
ume containers. Wipe the
rubber diaphragm or neck of
ampules with 70% alcohol.
7. For ampules, break neck at Avoid splattering of HEPA
score mark, pointing toward lter with liquids. Do not
side of workbench rather place hands between ltered
than HEPA lter. air source and critical site.
8. Attach previously sterilized Use rubber tubing. Make cer-
lter unit to vacuum sources. tain a trap ask is used to col-
Filter unit should contain lect ltrate overow. Falcon
47-mm, 0.45-m hydrophobic units may be stabilized by set-
edge membrane. ting in hold of blue plastic base.
9. Transfer the prescribed vol- See USP sterility test for
ume fromsample to upper number of samples and inocu-
chamber of lter unit. (a) Use a lum size. One syringe may
needle and syringe to withdraw be used for all samples since
the prescribedinoculumof prod- the samples will be pooled,
uct fromampules or vials. In- but a new sterile needle
sert needle through rubber clo- should be used for each vial
sure of vials or into opened or bottle.
ampules and withdrawpre-
scribed sample for test, or (b)
use a needle and transfer set to
transfer the prescribed volume
of solution fromlarge volume
containers. Insert spike of set
through rubber diaphragm.
376 Appendix II
Steps Comments
10. Wipe injection diaphragm of
lter unit with 70% alcohol.
11. Insert needle of syringe or Use proper aseptic tech-
transfer set through previ- nique. Be sure critical sites
ously asepticized diaphragm are bathed directly in LF air.
or administration set. A closed system is essential
to prevent drawing environ-
mental contaminants into up-
per chamber of lter unit.
12. Inject from syringe or apply Avoid direct injection on
vacuum to transfer pre- membrane as it may punc-
scribed volume of solution to ture the lter. Preferably, in-
be tested into upper chamber ject down side or into liquid
of the lter. layer above lter.
13. Apply vacuum to pull or Pull or push all solution
prime to push solution through lter.
through lter.
14. Repeat steps 610 until all
units have been tested.
15. When all solution has been Turn off vacuum carefully to
ltered, turn off vacuum and avoid reverse surge. Care
carefully remove top. must be taken to avoid acci-
dental contamination.
16. Aseptically pour 100 ml ster- To remove residual portions
ile diluting uid down inter- of product, rinse all surfaces
nal sides of chamber and efciently.
onto the lter. Replace top,
apply vacuum, and lter the
uid.
17. Repeat step 13 two more
times.
18. After all solution has been Exercise caution to avoid con-
ltered, turn off vacuum and tamination.
carefully remove top half of
lter assembly.
Appendix II 377
Steps Comments
19. Using sterile forceps and scis- Hold lter and cut over a
sors, remove membrane from sterile surface so that the
holder and cut into two membrane will not be acci-
halves. dentally contaminated if it
falls.
20. Place one half of the mem- Use sterile forceps to place
brane in a sterile tube of lter in culture media tubes.
SCD, the other half in a tube See USP for incubation times
of FTM, and incubate at pre- and temperatures.
scribed temperatures for the
specied time.
21. Include a positive and a nega- Inoculate one tube of each
tive control tube of each me- medium with a loopful of
dium. spore suspension or a paper
strip of B. subtilis as a posi-
tive control, plain medium as
a negative control.
22. Incubate the samples at ap- Incubate FTMat 3035C and
propriate temperatures and TSBat 2025C(roomtemper-
observe after 3 and 7 days ature [RT]). Longer incubation
for presence of microbial (e.g., 1014 days) may be re-
growth. quired at times to permit slow
growers to develop.
23. Thoroughly wash all equip- This is to remove residual
ment used. Make certain to product and media.
empty and clean vacuum
trap ask.
24. Return used equipment to
proper locations.
25. At the end of incubation pe-
riod observe samples for growth
andrecordresults of tests onap-
propriate report forms.
26. Sterilize used culture media
and clean tubes after incuba-
tion.
Appendix III
Aseptic Procedures at the
Laminar Flow Workbench*
Note: The laminar ow (LF) workbench with HEPA-ltered
air, when functioning properly, provides a Class 100 clean en-
vironment suitable for aseptic procedures. However, the pro-
cedures utilized must take advantage of the functional fea-
tures of the LF workbench in order not to compromise the
achievements possible therein.
1. The LF workbench should be located in a buffer area that
is clean and orderly, thereby enhancing the functional
efciency of the workbench.
2. At the beginning of each workday and each shift, and
when spillage occurs, the workbench surface should be
wiped thoroughly with a clean, nonlinting sponge damp-
ened with distilled water. The entire inside of the work-
bench should then be wiped with another clean, nonlint-
* Reprinted by permission of Dr. K. E. Avis, College of Pharmacy, University of Ten-
379
380 Appendix III
ing sponge dampened with a suitable disinfectant, such
as 70% alcohol.
3. The blower should be operated continuously. However,
should there be a long period of nonuse, the blower may
be turned off and the opening covered with a plastic cur-
tain or other shield. The blower then should be operated
for at least 30 minutes, and all the internal surfaces of
the hood should be cleaned thoroughly and wiped with a
disinfectant before use.
4. Trafc in the area of the workbench should be minimized
and controlled. The workbench should be shielded from
air currents that might overcome the air curtain and
carry contaminants into the work area.
5. Supplies entering the buffer area should be isolated in a
remote place until they can be decontaminated by remov-
ing outer packaging. That is, outer cartons and packag-
ing materials should not be brought near the workbench.
All supply items should be examined for defects prior to
being introduced into the asepetic work area.
6. Supplies to be utilized in the workbench should be decon-
taminated by wiping the outer surface with 70% alcohol
or other suitable surface disinfectants or by removing an
outer wrap at the edge of the workbench as the item is
introduced into the aseptic work area.
7. If the workbench is located in a non-aseptic area, such as
a hospital pharmacy, before approaching the workbench,
personnel must thoroughly scrub hands and arms with
a detergent followed by an appropriate skin antiseptic.
Each must then don a clean cap that provides complete
coverage of head hair and a clean, nonlinting, long-
sleeved coat with elastic or snaps at the wrist and, prefer-
ably, a solid front panel. A face mask must be worn if
there is no transparent barrier panel between the opera-
tors face and the aseptic work area or if the operator has
facial hair or an upper respiratory condition that pro-
motes sneezing and coughing.
8. After proper introduction of supply items into the aseptic
Appendix III 381
workbench, they are to be arranged in a manner such
that operations can take full advantage of the direction
of laminar air ow, that is, either vertical or horizontal.
Supply items within the workbench should be limited
to minimize clutter of the work area and provide ade-
quate space for critical operations. A clean pate of HEPA-
ltered air must be provided directly from the lter
source to the critical work site. No supplies and no move-
ment of the personnel should interpose a nonsterile item
or surface between the source of the clean air and the
critical work site. Therefore, no objects should be placed
horizontally behind the critical work site or above the
critical work site in a vertical laminar ow workbench.
Also, all work should be performed at least 6 inches
within the workbench to avoid drawing contamination in
from the outside.
9. All supply items should be arranged so that the work ow
will provide maximum efciency and order.
10. It should be noted that the hands are clean, but not ster-
ile. Therefore, all procedures should be performed in a
manner to minimize the risk of touch contamination. For
example, the outside barrel of a syringe may be touched
with the hands since it does not contact the solution, but
the plunger or needle should not be touched.
11. All rubber stoppers of vials and bottles and the neck of
ampules should be cleaned, preferably with 70% alcohol
and a nonlinting sponge, prior to the introduction of the
needle for removal or addition of drugs.
12. Avoid spraying solutions on the workbench screen and
lter.
13. After every admixture, the contents of the container
must be thoroughly mixed and should then be inspected
for the presence of particulate matter or evidence of an
incompatibility.
14. Filtration of solutions to remove particulate matter is
frequently necessary, particularly when admixtures
have been prepared. A small volume of solution may be
382 Appendix III
ltered by attaching an appropriate membrane lter to
the end of a syringe, using the plunger to force the liquid
through the lter. Note: To avoid rupture of the mem-
brane, force may be applied in one direction only through
the lter. When larger volumes of solutions must be l-
tered, this may be accomplished by means of an appro-
priate in-line lter and an evacuated container to draw
the solution through the lter or, preferably, by means
of a pressure tank of nitrogen, or other inert gas, to apply
pressure to the liquid in the container to force it through
the in-line lter. In the latter situation, the pressure
must be maintained lowenough to avoid the risk of explo-
sion of the solution container (usually a maximum of 10
12 psig). There are at least two disadvantages of the vac-
uum system as compared with the pressure system: (a)
any leakage draws contamination into the container and
system; (b) the vacuum may be lost, thereby stopping the
procedure.
15. The porosity of the appropriate membrane lter is deter-
mined by the objective of the ltration. To remove partic-
ulate matter, a 1-m porosity lter should be satisfac-
tory. To sterilize a solution, a 0.2-m lter would be
required.
16. The completed preparation should be provided with an
appropriate tamperproof cap or closure to assure the user
that the integrity of the container has been maintained
until the time of use.
17. The workbench should be cleaned with a clean sponge,
wet with distilled water, as often as necessary during the
workday and at the close of the workday. This should be
followed by wiping the area with a sponge with an appro-
priate disinfectant.
18. During procedures, used syringes, bottles, vials, and
other supplies should be removed, but with a minimum
of exit and reentry into the workbench.
1
Sterility Testing
INTRODUCTION
Sterility, or freedom from the presence of viable microorgan-
isms, is a strict, uncompromising requirement of an injectable
dosage form. Unlike interal administration, parenteral
(Greek, para enteron beside the intestine) administration
of drugs avoids many of the natural protective defenses of the
body. The injection of a product contaminated with living mi-
croorganisms would invite a multitude of complications to a
potentially immunocompromised patient.
When the termsterile appears on the label of a parenteral
product, it means that the batch or lot from which the sample
originated passed the requirements of the U.S. Pharmacopela
(USP) sterility test 71(or other national compendial steril-
ity test requirement). The USP sterility test provides an esti-
mate of the probable, not actual, sterility of a lot of articles.
The actual product administered to a patient has not been
tested for sterility. The sterility test is a destructive test; thus,
it is impossible to test every item for sterility. This presents
a major limitation of the sterility test. Sterility is based on
1
2 Chapter 1
the results of the testing of a small number of batch samples,
assuming that these samples are representative of every arti-
cle from the batch not tested for sterility. The answer to the
question whether the sample is representative of the whole
will always be uncertain. Furthermore, another limitation of
the sterility test is the nite frequency of accidental (or inad-
vertent) contamination of one or more samples during the per-
formance of the testing procedures. Regardless of the perfec-
tion attempted in the attitudes and techniques involved in
sterility testing, accidental contamination will occur with a
given percentage of tests conducted. The use of barrier isola-
tion technology (compared to use of a conventional clean room)
by the pharmaceutical industry has greatly reduced the
chance of accidental contamination that can yield a false posi-
tive sterility test.
In light of these and other limitations of the USP sterility
test, why is it still a requirement of and enforced by the Food
and Drug Administration (FDA) and other regulatory agen-
cies? The most important and obvious reason is to provide
some means, albeit small, of end-product testing to protect the
consumer from administration of a contaminated injectable
product.
An exception to end-product sterility testing involves ter-
minally sterilized large-volume parenterals, which have been
exposed to sterilization conditions experimentally validated to
assure product sterility well beyond the capability of sterility
testing to detect contamination, while products that are termi-
nally sterilized usually have a sterility assurance level (SAL)
of at least 10
6
. Release of products without end-product steril-
ity testing but based on validation of the sterilization process
is called parametric release (this will be discussed at the end
of this chapter).
While the sterility test does not assure sterility of every
article, it does provide the FDA, the manufacturer, and the
user with some end-point check that a representative sample
of the batch does not disclose the existence of a high proportion
of contaminated units in a lot or batch. End-product sterility
testing also presents a reliable means of checking the sterility
Sterility Testing 3
of a product that has been sterilized by marginal sterilization
processes such as aseptic ltration. More discussion of this
controversial subject is presented below.
Even if more reliable sterilization methods are used, ste-
rility testing provides an additional means of checking that all
facets of the sterilization process were achieved. For example,
although steam sterilization is the most reliable sterilization
process known, improper loading of the autoclave might pre-
vent adequate steam penetration of some of the product con-
tainers, in the batch. A statistically sound sampling procedure
(again, a necessary assumption of the sterility test) will select
one or more of those improperly exposed containers, and the
sterility test will show contamination. Nevertheless, it must
be recognized, as it is by the USP, that the sterility test was
not designed to ensure product sterility or sterilization process
efcacy (1). It simply is a procedure used for sterility control
and assurance, along with many other procedures used in
manufacturing to assure the sterility of a product.
This chapter will present a thorough and practical analy-
sis of the ofcial testing requirements for sterility, their ad-
vantages and limitations, and current adjunct processes and
controls to aid in the proper performance and valid interpreta-
tion of the sterility test. Also, appropriate focus is placed on
the current issues of sterility testing, including retesting of
initial test failures, newtechnology, and sterility testing in the
hospital pharmacy. Other review articles on sterility testing
include those by Bowman (2), Borick and Borick (3), Beloian
(4), Outschoorn (5), and Olson (6). The reader should also be
familiar with a published survey of sterility test practices con-
ducted by the Parenteral Drug Association in 1987 (7). This
survey can supplement much of the material emphasized in
this chapter.
STERILITY AND STERILITY TEST REGULATIONS
Sterility is the most important and absolutely essential char-
acteristic of a parenteral product. Sterility means the com-
plete absence of all viable microorganisms. It is an absolute
4 Chapter 1
term; that is, a product is either sterile or not sterile. Building
sterility into a product through meticulously validated clean-
ing, ltration, and sterilization procedures is more preferable
than testing for sterility of a product subjected to marginal
or inadequate production processes. The sterility test should
never be employed as an evaluation of the sterilization pro-
cess. Sterility and quality cannot be tested into a product; they
can only be components of controlled processes throughout the
production sequence (7). The sterility test, however, should be
employed as the last of several checkpoints in reaching a con-
clusion that the production process has removed or destroyed
all living microorganisms in the product (2).
The USP chapter 1 on injections states that prepara-
tions for injection meet the requirements under Sterility
Tests. After meeting these requirementsthat is, all media
vessels incubated with product sample reveal no evidence of
microbial growth (turbidity)the tested product may be
judged to meet the requirements of the test. If evidence for
microbial growth is found, the material tested has failed to
meet the requirements of the test for sterility. Retesting is
only allowed if there is unequivocal proof that the failed result
was due to operator or accidental contamination. The FDAhas
stringent requirements for sterility retesting.
Evidence for microbial growth is determined by visual
evaluation of a vessel containing the product sample in the
proper volume and composition of nutrient solution. Provided
that the growth conditions are optimalproper nutrients, pH,
temperature, atmosphere, sufcient incubation time, and so
ona single microbial cell will grow by geometric progres-
sion* until the number of microbial cells and their metabolic
products exceed the solubility capability of the culture me-
dium. Manifestation of this overgrowth is visualized by the
* Microbial growth may be characterized by the equation N 2
gt
, where N is the
number of microbial cells, g is the number of generations or replications, and t is the
time period during growth. For example, a cell that replicates once every 30 minutes
will, after 10 hours, grow to 2
210
2,097,152 cells!
Sterility Testing 5
appearance of a cloudy or turbid solution of culture medium.
A noxious odor may also accompany the turbid appearance of
the contaminated medium. The sterility test is a failure when
a product generates turbidity in a vessel of culture medium
while the same lot of medium without the product sample
shows no appearance of turbidity.
Parenteral drug administration was a routine practice in
the early 1900s. For example, insulin was discovered in 1921
and was, as it is today, administered by subcutaneous injec-
tion. Yet, the rst ofcial compendial requirement of sterility
testing of drugs administered by the parenteral route did not
appear until 1932 in the British Pharmacopoeia. Sterility
tests were then introduced in the 11th edition of the USP and
in the sixth edition of the National Formulary (NF) in 1936.
During the past 56 years and longer, signicant changes and
improvements have occurred in the ofcial sterility test re-
quirements; a summary appears in Table 1.1.
Congress passed the Federal Food, Drug, and Cosmetic
(FD & C) act in 1938, permitting the FDA to enforce the act.
The act recognized the USP and the NF as ofcial compendia
to describe the standards of safety, identity, strength, quality,
and purity of drugs and the drug dosage form. In 1975, the
two compendia were unied. In 1976, the FD & C Act was
amended to recognize medical devices as entities to be in-
cluded in the compendia. Thus, all drug and device products
that bypass the gastrointestinal tract on administration to a
human being or animal must pass the USP sterility test; this
requirement is strictly enforced by the FDA.
Besides the USP/NF ofcial compendia, regulations also
exist for two specic groups of pharmaceuticals, the biologics
(vaccines, serums, toxins, antitoxins, and blood products) and
antibiotics. Sterility tests for biologics and antibiotics are de-
scribed under Title 21 of the Code of Federal Regulations
(8).
In 1978, the nal approved regulations of the FDA-
authored current good manufacturing procedures (CGMPs)
was published. Sterility testing was briey mentioned under
6
C
h
a
p
t
e
r
1
Table 1.1 Summary of Changes and Improvements in the USP Requirements for Sterility Testing
Year USP edition Change or improvement
1936 11th First year sterility test appeared, applied only to sterile liquids.
1942 12th Aerobic sterility test in sterile solids and liquids.
Procedures for inactivation of certain preservatives.
1945 13th Fluid thioglycollate medium (FTM) introduced for recovery of aerobic and anaer-
obic bacteria.
Honey medium introduced for recovery of molds and yeasts.
Brief description of laboratory area and training of personnel to perform steril-
ity tests.
1950 14th Incubation temperature of FTM lowered from 37C to 3235C.
Sabouraud liquid medium (modied) replaced honey medium.
1955 15th USP Fluid Sabouraud medium replaced the modied Sabouraud medium.
1970 18th Soybean-casein digest medium replaced.
Sabouraud medium.
Membrane ltration sterility test introduced.
Guidelines included for specic use of biological indicators.
Expanded sections on describing the area, personnel training, and techniques
for performing sterility tests.
1975 19th Established separate section for membrane ltration procedures.
Included test procedure for large-volume solutions (100 ml).
1980 20th Section introduced on growth promotion testing using specic indicator microor-
ganisms.
Section introduced on sterility testing of prelled disposable syringes.
Provided guidelines on rst and second retests of suspected false-positive tests.
S
t
e
r
i
l
i
t
y
T
e
s
t
i
n
g
7
General expansion and/or elaboration of sections of bacteriostasis, sterility test-
ing of devices, and sterilization.
Sterilization section contained statement of F
0
8 minutes for steam-sterilized
articles and D values provided for biological indicators.
1985 21st Sections on biological indicator paper strips for dry heat sterilization, ethylene
oxide sterilization, and steam sterilization.
Stricter requirements for repeating failed sterility tests.
Section on basic principles of process validation in the sterility assurance of
compendial articles.
Expansion of information on sterilization by ionizing radiation and ltration.
Section on denition of a lot for sterility test purposes.
Reorganization and expansion of section on performance, observation, and inter-
pretation of sterility test results.
Deletion of procedures for sterility testing of sutures and petrolatum gauze.
1990 22nd Expansion of bacteriostasis and fungistasis section on use of membrane ltra-
tion.
Further guidance for membrane ltration testing of product having inherent
bacteriostatic properties.
New section on membrane ltration testing procedure for lterable solids.
2000 24th Addition of Pseudomonas aeruginosa and Staphylococcus aureus to the
bacteriostasis/fungistasis test. New guidance on the use of isolators and steril-
ity testing areas. Addition of the 14 day incubation period for products which
are not terminally sterilized by moist heat.
8 Chapter 1
Sect. 211.167, For each batch purporting to be sterile, there
shall be appropriate tests to determine conformance to such
requirements. To elaborate on this requirement and to ad-
dress more specic issues confronted by both industry and the
FDA in manufacturing and control of aseptically produced
drug products, the FDA published its Guidelines on Sterile
Drug Products Produced by Aseptic Processing in 1987.
These guidelines as they relate to sterility testing are covered
in detail below.
SAMPLING FOR STERILITY TESTING
In pharmaceutical manufacture, the sterility of a parenteral
product lot is checked by a statistically valid sampling proce-
dure. After years of experience, most manufacturers of paren-
teral products will sterility test 10 to 20 units of product per
lot. The number of units tested may be doubled when the deliv-
erable volume is 1 ml or less. The number of units sampled
depends on the number of units in the batch, the volume of
liquid per container, the method of sterilization, the use of a
biological indicator system, and the good manufacturing prac-
tice requirements of the regulatory agency for the particular
product. For example, if the batch size is greater than 500 arti-
cles, a minimum of 20 units is sampled. If the nal batch size
is between 100 and 500 articles, then no fewer than 10 of the
articles are sterility tested, although there are minimum re-
quirements for sterility testing of biologics. For large-volume
parenteral (LVP) products (volume 100 ml per container),
at least 2% of the batch or 10 containers, whichever is less, is
sampled. Sampling requirements as specied in the USP and
European Pharmacopeia (EP) Sterility Test Section are sum-
marized in Table 1.2.
Correct statistical sampling represents a difcult, yet vi-
tal, aspect of sterility testing. Realizing that the parenteral
product being used by the patient has itself not been tested
for sterility, it is absolutely essential that the sampling proce-
dure be as valid and representative of the whole batch as pos-
S
t
e
r
i
l
i
t
y
T
e
s
t
i
n
g
9
Table 1.2 Minimum Number of Units Required per Medium for Performance of the USP Sterility
Test as a Function of Volume per Test Unit (Quantities for Liquid Articles)
Minimum volume of each medium
Used for membrane
or half membrane
Minimum volume Used for direct representing total
taken from each transfer of volume volume from the
Container container for taken from each appropriate number No. of containers
content (ml) each medium container (ml) of containers (ml) per medium
Less than 10
a
1 ml, or entire 15 100 20 (40 if each
contents if less does not con-
than 1 ml tain sufcient
volume for both
media)
10 to less than 5 ml 40 100 20
50
a
50 to less than 10 ml 80 100 20
100
a
50 to less than Entire contents 100 10
100, intended
for intravenous
administration
b
100 to 500
a
Entire contents 100 10
Antibiotics 1 ml 100 10
(liquid)
a
Intended for multiple dose or nonintravenous use.
b
Intended for single dose or intravenous use.
10 Chapter 1
sible. Realistically, this presents an impossible principle to
prove.
Pharmaceutical quality control departments employ
sampling plans called acceptance sampling for many quality
control testing procedures that are not amenable to 100%nal
testing. Acceptance sampling in sterility testing is based on
the establishment of operating characteristic (OC) curves,
which are plots of probability versus percentage contamina-
tion. Operating characteristic curves for sample sizes of 10
and 20 units are shown in Figs. 1.1 and 1.2, respectively (9).
These curves are drawn from a series of government-spon-
sored sampling plans called MIL-STD-414 (10). The shape of
the curve depends on ve criteria:
1. An acceptable quality level (AQL), which is the highest
percentage of defective (nonsterile) units that is accept-
able
2. An unacceptable quality level (UQL), which is the percent-
age of nonsterile units for which there is a low probability
of acceptance
3. The alpha () factor, which is the probability of rejecting
a good (sterile) batch
4. The beta () error, which is the probability of accepting a
bad (nonsterile) batch
5. The sample size
With all criteria (1) through (5) being constant, the slope
of the OC curve will become steeper as the sample size is in-
creased. Similarly, with the criteria being constant, the slope
of the curve will become steeper as the AQL is decreased or as
the UQLis decreased. An example of an OCcurve for sampling
plans at AQL 1% for different sample sizes is seen in Fig.
1.3 (11). At a given AQL level, the larger the sample size, the
greater is the probability of accepting a sterile lot and re-
jecting a nonsterile lot. Each pharmaceutical manufacturer
for each type of parenteral product assumes a given AQL or
rate of contamination, thus xing the point of reference on the
abscissa of the OC curve.
Sterility Testing 11
Fig. 1.1 Operation characteristic curves for a sample size of 10
units; n sample size; N lot size. (From Ref. 9.)
Sampling plans and concomitant OC curves are prepared
on the assumption that the samples are selected at random.
By random sampling, it is inferred that any one of the re-
maining uninspected units of the same lot of product has an
equal chance of being selected (12). This is not always easily
accomplished. Random sampling often is inconvenient and
may not be appreciated by production workers responsible for
many other important duties during the production process.
Random samples are optimally selected every k
th
unit, where
12 Chapter 1
Fig. 1.2 Operating characteristic curves from a sample size of 20
units; n sample size, N lot size (From Ref. 9.)
k the total units in the batch per the number of samples
required. For example, if the batch size of an aseptically lled
product is 10,000 units and 20 samples are required for the
sterility test, then samples are taken every 500 units includ-
ing the rst and last unit lled.*
* Some manufacturers also take sterility test samples immediately after a halted
sterile (aseptic) production process has been re-started.
Fig. 1.3 Operating characteristic curve for sampling plans at an
acceptable quality level of 1% for different sample sizes N. (From
Ref. 11.)
Additional discussion of sampling with regard to its lim-
iting the interpretation of the results of the sterility test is
presented in this chapter in the section, Limitations of the
USP/NF Referee Sterility Test.
A major consideration in sampling for sterility testing is
proper treatment of the package systemto prevent contamina-
tion of the sample when it is taken out of the package for test-
ing. For example, parenteral products packaged in ampules,
vials, or bottles must be aseptically sampled using sterile ma-
terials and aseptic techniques. The neck of the ampule or the
surface of the rubber closure must be disinfected with a liquid
disinfectant solution before breaking the ampule or penetrat-
ing the closure with a needle. Special procedures must be im-
plemented to sample products contained in aluminum foil, pa-
per, or plastic outer bags. For example, bulk solid chemicals
sterilized by ethylene oxide prior to aseptic compounding are
contained in gas-permeable paper or plastic bags. The chemi-
cals must be sampled by tearing open the package, which is
not easy to do because of the potential for accidental contami-
nation. Sutures are contained in glass or aluminum foil enclo-
14 Chapter 1
sures that must be disinfected before the product is removed.
Sampling of devices without contaminating the sample also is
a very difcult procedure to accomplish. Although the package
may be designed to maintain the sterility of the product in-
denitely, it is obviously of no value if the inner contents can-
not be removed without contaminating the product and in-
terfering with the performance of certain essential tests (3).
CULTURE MEDIA
The USP and EP describe two primary types of culture media
to be used in the sterility testing of parenteral products. One
type is called uid thioglycollate medium (FTM), which was
introduced by Brewer (13) in 1949. The formulation ingredi-
ents of FTM and their basic purpose in the medium are listed
in Table 1.3.
FTM provides both aerobic and anaerobic environments
within the same medium. Thioglycollate and L-cysteine are
antioxidants or reducing agents that maintain anaerobiasis in
the lower levels of the culture tube. FTM solution has a two-
Table 1.3 Ingredients of Fluid Thioglycollate medium and Their
Purpose
L-Cysteine 0.5 g Antioxidant
Agar, granulated (moisture 0.75 g Nutrient and viscosity
content 15%) inducer
Sodium chloride 2.5 g Isotonic agent
Dextrose 5.5 g Nutrient
Yeast extract 5.0 g Nutrient
Pancreatic digest of casein 15.0 g Nutrient
Sodium thioglycollate or 0.5 g Antioxidant
thioglycollic acid 0.3 ml
Resazurin sodium solution 1.0 ml Oxidation indicator
(1: 1000), freshly pre-
pared
Puried water QS 1000 ml
pH after sterilization 7.1 0.2.
color appearance. The pinkish color of the top part of the solu-
tion is indicative of the presence of resazurin sodium, an oxy-
gen-sensitive indicator. The pink color should consume no
more than one-third of the medium volume. Because of the
need for two environments in the same test tube or container,
the ratio of the surface to the depth of the medium is very
important. To provide adequate depth for oxygen penetration,
a 15-ml volume of FTM must be contained in a test tube with
dimensions of 20 150 mm. A 40-ml volume of FTM is to be
contained in 25 200 mm test tubes, and 75100 ml FTM in
38 200 mm test tubes.
Devices containing tubes with small lumina are sterility
tested using an alternate thioglycollate medium in which the
agar and resazurin sodium are deleted. The same medium is
used for turbid or viscous parenterals. Without the agar, the
medium will not interfere with the viscosity of the product or
be as resistant in lling small lumina. Since the medium will
be turbid, the presence of a color indicator would not be seen
anyway. For oily products, FTM is slightly modied by the ad-
dition of 1 ml Polysorbate 80 to 1 liter of the media. Polysor-
bate 80 serves as an emulsifying agent to permit adequate dis-
persal of a lipophilic product in a hydrophilic growth medium.
FTM is an excellent medium for the detection of bacterial
contamination. Thioglycollate also has the advantage of neu-
tralizing the bacteriostatic properties of the mercurial preser-
vatives. One disadvantage of FTM is that it will not support
the growth of Bacillus subtilis spores entrapped in solids or
material that locates itself in the anaerobic lower portion of
the medium (14). Bacillus subtilis spores require an environ-
ment of high surface tension for normal growth.
The other primary USP/NF culture medium for the ste-
rility testing of parenterals is called soybean-casein digest
(SCD) or trypticase soy broth (TSB) medium. The formulation
ingredients and their purpose in TSB are shown in Table 1.4.
TSB has a slightly higher pH (7.3 0.2) than FTM (7.1
0.2). TSB replaced Sabouraud medium in the 19th edition of
the USP (1970) because TSB was found from experience to
16 Chapter 1
Table 1.4 Ingredients of Trypticase Soy Broth and Their
Purpose
Pancreatic digest of casein 17.0 g Nutrient
Papaic digest of soybean meal 3.0 g Nutrient
Sodium chloride 5.0 g Isotonic agent
Dibasic potassium phosphate 2.5 g Buffer
Dextrose 2.5 g Nutrient
Puried water
a
QS 1000 ml
pH after sterilization 7.3 0.2.
a
Distilled or deionized water can be used instead of puried water.
be a better medium. It possesses a higher pH and thus was
considered a better nutrient for fungal contaminants (15).
Fluid Sabouraud, designed to inhibit certain bacteria, was
successful in promoting the growth of molds, fungi, and other
saprophytes requiring high dextrose content and lowpH. TSB,
however, promotes growth of fungi and bacteria and is also
considered a better medium for than FTM slow-growing aero-
bic microorganisms.
Other media have been proposed to replace or to substi-
tute for FTM and/or TSB. Abdou (16) found that a dithionite-
thioglycollate broth and a peptone liver digest medium were
superior to FTM and TSB in growing various strains of bacte-
ria, yeasts, and molds. Concentrated brain heart infusion
broth has been suggested as an alternative to FTM and TSB
when large-volume parenterals are directly inoculated with
culture medium. Table 1.5 lists the formulas of eleven media
and reagents potentially used in the sterility testing of paren-
teral products. While these and other media might be appro-
priate for certain products or situations, it is highly unlikely
that TSB or FTM will be replaced as ofcial USP or EP steril-
ity test media.
Culture media may be purchased in either the dehy-
drated state or the ready-to-use uid state. Dehydrated media
are less expensive and have a longer shelf life. Strict adher-
ence to the expiration date on the label of premixed culture
Table 1.5 Media and Reagents Potentially Used in Performing
the USP/NF Sterility Tests
a
USP uid thioglycollate (thio) medium. Use BBL
11260 or Difco 0256.
Trypticase Peptone (BBL) or Bacto-Casitone (Difco) 15.0g
L-Cysteine 0.5 g
Dextrose (anhydrous) 5.0 g
Yeast extract 5.0 g
Sodium chloride 2.5 g
Sodium thioglycollate 0.5 g
Resazurin (1: 1000) 1.0 ml
Agar 0.75 g
Puried water
b
1.0 liter
Final pH 7.1
USP soybean-casein digest medium. Use Trypti-
case Soy Broth (BBL 11768) or Tryptic Soy Broth
(Difco 0370).
Trypticase Peptone (BBL) or Bacto-Tryptone (Difco) 17.0 g
Phytone Peptone (BBL) or Bacto-Soytone (Difco) 3.0 g
Dipotassium phosphate 2.5 g
Dextrose 2.5 g
Puried water
b
1.0 liter
Final pH 7.3
Polysorbate 80. A suitable grade is TWEEN 80,
available from Atlas Chemicals Division, ICI
Americas Incorporated.
Brain heart infusion. Use BBL 11059 or Difco
0037.
Calf brain, infusion from 200.0 g
Beef heart, infusion from 250.0 g
Gelysate Peptone (BBL) or Proteose Peptone 10.0 g
(Difco)
Disodium phosphate 2.5 g
Dextrose 2.0 g
Distilled water 1.0 liter
Final pH 7.4
18 Chapter 1
Table 1.5 Continued
Sporulating agar medium. Use AK Agar No. 2
(Sporulating Agar) (BBL 10912) or Sporulating
Agar (Difco 0582).
Gelysate Peptone (BBL) or Bacto-Peptone (Difco) 6.0 g
Trypticase Peptone (BBL) or Bacto-Casitone (Difco) 4.0 g
Yeast extract 3.0 g
Beef extract 1.5 g
Dextrose 1.0 g
Agar 15.0 g
Manganous sulfate 0.3 g
Final pH 6.5
Saline TS, Sterile (USP).
Puried water 1.0 liter
Sabouraud dextrose agar medium. Use BBL
11584 or Difco 0109.
Dextrose 40.0 g
Polypeptone (BBL) or Neopeptone (Difco) 10.0 g
Agar 15.0 g
Final pH 5.6
USP soybean-casein digest agar medium. Use
Trypticase Soy Agar (BBL 11043) or Tryptic Soy
Agar (Difco 0369).
Trypticase Peptone (BBL) or Bacto-Tryptone (Difco) 15.0 g
Phytone Peptone (BBL) or Bacto-Soytone (Difco) 5.0 g
Agar 15.0 g
Puried water
b
1.0 liter
Final pH 7.3
Fluid Sabouraud Medium. Use Sabouraud Liquid
Medium (Difco 0382).
Dextrose 20.0 g
Polypeptone Peptone (BBL) or Neopeptone (Difco) 10.0 g
Final pH 5.7
Table 1.5 Continued
USP antibiotic (agar) medium 1. Use BBL 10937
or Difco 0263.
Gelysate peptone (BBL) or Bacto-Peptone (Difco) 6.0 g
Trypticase Peptone (BBL) or Bacto-Casitone (Difco) 4.0 g
Yeast extract 3.0 g
Beef extract 1.5 g
Dextrose 1.0 g
Agar 15.0 g
Puried water
b
1.0 liter
Final pH 6.6
Potato dextrose agar medium
Potato agar 15.0 g
Glucose 20.0 g
Distilled water q.s. 1.0 liter
pH 5.6 0.2
a
Sterilize in an autoclave at 121C (15 lb pressure) for 15 minutes unless
otherwise indicated. If commercial preparations are not available, equiva-
lent preparations may be used.
b
Distilled or deionized water can be used instead of puried water.
media tubes must be obeyed, provided that the proper storage
conditions (usually refrigeration) have been met.
Preparation of sterile uid culture media from dehy-
drated media is a relatively simple process. The label of each
container of medium describes the procedure for preparation.
Basically, the procedure involves (a) weighing the appropriate
amount of medium per liter of uid desired, (b) adding water
to the compounding vessel to the desired volume, (c) slowly
adding the culture medium while stirring the solution, (d)
applying heat and stirring until the medium is completely dis-
solved, and (e) sterilizing the medium in bulk or after lling
into test tubes or other containers by steam heat under pres-
sure by a validated sterilization cycle. Before discarding con-
taminated culture media, they must be again sterilized by
steam under pressure before pouring the uid into a drainage
system and washing the containers.
20 Chapter 1
Table 1.6 Formulations of Various Diluting Fluids
Used with the Membrane Filtration Test Method
Diluting Fluid A
Peptic digest of animal tissue 1.0 g
pH 7.1 0.2
Diluting Fluid D
Polysorbate 80 1.0 ml
pH 7.1 0.2
Diluting Fluid A modied
Ascorbic acid 10.0 g
Diluting Fluid E
Isopropyl myristate 100.0 ml
Water extract pH not less than 6.5
Medium K
Beef extract 3.0 g
Polysorbate 80 10.0 g
Distilled water q.s. 1.0 liter
pH 6.9 0.2
When membrane ltration is used for the sterility test,
a diluting uid must be used to rinse the ltration assembly
to ensure that no microbial cells remain anywhere but on the
lter surface. The diluting uid may also be used to dissolve
a sterile solid prior to ltration. Some examples of diluting
uid formulas are listed in Table 1.6. Diluting uids are in-
tended to minimize the destruction of small populations of
vegetative cells during the pooling, solubilizing, and ltering
of sterile pharmaceutical products (17).
TIME AND TEMPERATURE OF INCUBATION
No ideal incubation time and temperature condition exists for
the harvesting of all microorganisms. Most organisms grow
more rapidly at 37C than at lower temperatures. However,
a temperature of about 23C may reveal the presence of some
organisms that might remain undetected if incubations were
done at higher temperatures (18). Pittman and Feeley (19)
demonstrated that temperatures of 22C and 30C were more
favorable for the recovery of yeasts and fungi in FTM than a
temperature of 35C. The Division of Biologics Standards of
the National Institutes of Health discovered that a pseudomo-
nad contaminant in plasma grew in FTM at 25C, but was
killed at 35C (2). As a result of this nding, the incubation
temperature range of FTM was lowered from 3235C to 30
35C as required by the USP/NF (20th edition).
The current time and temperature incubation require-
ments of the USP sterility test are found in Table 1.7. Incuba-
tion in TSB is accomplished at 2025C because of favorable
growth of fungal and slow-growing aerobic contaminants at
this temperature range. The time of incubation for sterility
testing by membrane ltration (MF) is 7 or 14 days.
Findings by Bathgate et al., suggest that a 7-day incuba-
tion period is inappropriate for detecting a potentially contam-
inated unit. The authors suggest that at least 14 days is re-
quired to signicantly increase the chance of detecting most
contamination in the sterility test and cite several examples
for which certain tests were negative after 7 days of incuba-
tion. Some contaminants did not begin to show signs of visible
growth until the 10th day of incubation. The article reports
additional problems when ethylene oxide was used for steril-
ization or when an antimicrobial preservative was employed
in the drug product. The data in this article also suggest that
there is no signicant advantage of the MF test when com-
pared to the direct transfer (DT) test (105).
Others disagree with Bathgate et al. (105), citing the al-
ready low chance of the compendial sterility test to detect a
contaminated unit in a batch. These authors cite many rea-
sons why a 7 day incubation period is adequate to detect con-
tamination in a sterility test (106).
No doubt the debate to continue the sterility test will con-
tinue for years. The time and temperature incubation require-
2
2
C
h
a
p
t
e
r
1
Table 1.7 Time and Temperature Incubation Requirements of the USP Sterility Test
Time Temperature
Medium Test procedure (days)
a
(C) Sterilization method
FTM Direct transfer 14 3035 Steam or aseptic process
Membrane ltration 100 ml 7 3035 Terminal moist heat sterilization
Membrane ltration 100 ml 14 3035 Aseptic process
TSB Direct transfer 14 2025 Steam or aseptic process
Membrane ltration 100 ml 7 2025 Terminal moist heat sterilization
a
Minimum number of incubation days. Additional incubation time may be required if the product is conducive to
producing a slow-growing contaminant.
S
t
e
r
i
l
i
t
y
T
e
s
t
i
n
g
2
3
Table 1.8 Time and Temperature Incubation Requirements of the EP Sterility Test
Time Temperature
Medium Test procedure (days)
a
(C) Sterilization method
FTM Direct transfer 21 (14 7) 3035 Steam or aseptic process
Membrane ltration 100 ml 7
b
3035 Terminal moist heat sterilization
TSB Direct transfer 21 (14 7) 2025 Steam or aseptic process
Membrane ltration 100 ml 7
b
2025 Terminal moist heat sterilization
a
Minimum number of incubation days. Additional incubation time may be required if the product is conducive to
producing a slow-growing contaminant.
b
A 7-day incubation period is permissible only when authorized or dictated in the European Medicines Evaluation
Agency (EMEA) submission. In general, a 14-day incubation period is required for all products that are required to
meet the EP sterility test.
24 Chapter 1
ments of the USP and EP tests are summarized in the Tables
1.7 and 1.8, respectively.
The incubation time requirements of the sterility test
must be sufciently long to account for the variable lag time
characteristic of the growth curve of most microbial forms. A
typical growth cycle for bacteria is seen in Fig. 1.4. At the be-
ginning of the cycle, corresponding to the time at which the
test sample is combined with the culture medium, there exists
a lag time phase. The length of this time depends on the rapid-
ity of the microbial cell to adapt to its new environment. Usu-
ally, the lag phase lasts no longer than a few hours, but the
possibility is always present that a resistant spore form, a
slow-growing contaminant, or a microorganism with an extra
Fig. 1.4 Typical growth-and-death cycle for bacteria.
long lag phase owing to damage caused in the sterilization pro-
cess may be part of the test sample. Sufcient incubation time
must be allowed for the microbial form to overcome its own
resistance to grow in the FTM or TSB environment. However,
once the lag phase is completed, the growth phase is exponen-
tial. Most contaminated samples will show evidence of con-
tamination within 24 to 48 hours due to the meteoric growth
of microorganisms.
For the direct transfer method, it is possible that a physi-
ochemical incompatibility between the product inoculum and
the culture medium might exist, resulting in a precipitate or
turbid reaction not indicative of microbial growth. Should this
occur, the appropriate action to take is to transfer an aliquot
(usually 1 ml) of the suspension to fresh culture medium on
the days 3 to 7 (for the USP test) and day 14 (for the EP test)
after the test was started and incubate both the original and
the new media for a total of, but not less than, 7 additional
days. Thus, in all cases, the original and aliquot vessels would
be incubated for 14 days (for the USP test) or 21 days (for the
EP test). Incubation time and temperature requirements are
shown in Table 1.7 and 1.8 for the USP and EP sterility tests,
respectively.
Incubation time and temperature requirements for steril-
ity tests conducted under the auspices of various authorities
are basically similar to those of the USP presented in Table
1.7, although subtle differences in the sterility test do exist
for the EP sterility test (see Table 1.8). The importance of time
of sampling in hospital intravenous admixture sterility testing
was discovered by DeChant et al. (20). They found that no
more than 1 hour should transpire between preparation of in-
travenous admixtures and sampling of the admixture for con-
ducting a sterility test. If longer time periods are permitted,
microorganisms, if introduced during the admixture prepara-
tion period, may be inhibited from reproducing because of bac-
tericidal activity of certain intravenous solutions, such as dex-
trose 5% in water.
26 Chapter 1
STERILITY TEST METHODS
The USP and EP sterility tests specify two basic methods for
performing sterility tests, the direct transfer (DT) or direct in-
oculation method and the MF method, with a statement that
the latter, when feasible, is the method of choice. In fact, in
some cases, membrane ltration may be the only possible
choice. Suggested standard operating procedures for per-
forming both methods are given at the end of this book as Ap-
pendices I and II.
Direct Transfer Method
The DT method is the more traditional sterility test method.
Basically, the DT method involves three steps:
1. Aseptically opening each sample container froma recently
sterilized batch of product
2. Using a sterile syringe and needle to withdraw the re-
quired volume of sample for both media fromthe container
3. Injecting one-half of the required volume sample into a
test tube containing the required volume of FTM and the
other half volume of sample into a second test tube con-
taining the required volume of TSB
The DT method is simple in theory, but difcult in prac-
tice. The technician performing the DT test must have excel-
lent physical dexterity and the proper mental attitude about
the concern for maintaining asepsis. The demand for repeti-
tion in opening containers, sampling, transferring, and mixing
can potentially cause fatigue and boredom, with a subsequent
deterioration in operator technique and concern. As this oc-
curs, the incidence of accidental product sterility test contami-
nation will increase.
The USP and EP tests require a minimumvolume of sam-
ple per container volume to be transferred to a minimum vol-
ume of each culture medium. Table 1.9 lists these volume re-
quirements. The sample volume must be a sufcient
representation of the entire container volume and the volume,
Table 1.9 Volume Requirements of the Direct Transfer Sterility
Test
Container Minimum volume Minimum volume
content (ml) of product (ml) of medium (ml)
10 or less 1 (or total contents if less 15
than 1 ml)
1050 5 40
50100 10 80
100500 One-half contents N/A
500 500 N/A
Antibiotics (liquid) 1 ml N/A
N/A, not applicable.
of medium must be sufcient to promote and expedite micro-
bial growth, if present. Adequate mixing between the sample
inoculum and the culture medium must take place to max-
imize interaction and facilitate microbial growth.
Membrane Filtration Method
The MF sterility test became ofcial in the 18th edition of the
USP in 1970. It has since become the more popular and widely
used method over the DT method and, when feasible for phar-
macopeial articles, should be preferred. Specic application of
the MF sterility test method has been the subject of many pub-
lications, for example, those concerned with the sterility test-
ing of antibiotics (17), insulin (21), and LVPs (22).
The successful employment of this technique requires
more skill and knowledge than that required for the DT
method. Five basic steps are involved in the use of the MF
sterility test method:
1. The lter unit (Fig. 1.5) must be properly assembled and
sterilized prior to use.
2. The contents of the prescribed number of units are trans-
ferred to the lter assembly under strict aseptic condi-
tions.
28 Chapter 1
Fig. 1.5 Set-up of the classic membrane ltration sterility test
apparatus. (Courtesy of Eli Lilly Co., Indianapolis, Indiana.)
3. The contents are ltered with the aid of a vacuum or pres-
sure differential system.
4. The membrane is removed aseptically and cut in half.*
5. One-half of the membrane is placed in a suitable volume
(usually 100 ml) of FTM, and the other membrane half is
placed in an equal volume of TSB.
A suitable membrane lter unit consists of an assembly that
facilitates the aseptic handling of the test articles and allows
the processed membrane to be removed aseptically for trans-
fer to appropriate media or an assembly by which sterile me-
dia can be added to the sealed lter and the membrane incu-
bated in situ. A membrane suitable for sterility testing has a
rating of 0.45 m and a diameter of approximately 47 mm.
* The USP gives the option of using two whole membranes, one for each medium,
or cutting a single membrane.
These membranes have hydrophobic edges or low product-
binding characteristics that minimize inhibitory product-resi-
due, and it is this residue that interferes with requirements
of the validation test for bacteriostasis and fungistasis. For
products that do not contain inhibitory substances, mem-
branes without hydrophobic edges can be used, but should be
wetted prior to testing.
The cleaning, assembly, sterilization, and nal connec-
tions involved in the preparation of the membrane ltration
equipment are described in Appendix III. Complete descrip-
tion and application of membrane sterility test methods for
antibiotics, nonantibiotics, and ophthalmics may be studied
using the Millipore Application Manual AM201 (Millipore
Corp., Bedford, MA 01730).
The membrane ltration technique has been further de-
veloped by the Millipore Corporation, and the Steritest
sys-
tem has become widely used when the MF technique is em-
ployed. The Steritest system is essentially the same as the
conventional MF system, but greatly reduces the chance of ac-
cidental contamination of the test by keeping the sterility test
system as closed as possible.
The MF method offers at least ve advantages over the
use of the DT method:
1. Greater sensitivity (23).
2. The antimicrobial agent and other antimicrobial solutes
in the product sample can be eliminated by rinsing prior
to transferring the lter into test tubes of media, thereby
minimizing the incidence of false-negative test results.
3. The entire contents of containers can be tested, providing
a real advantage in the sterility testing of large-volume
parenterals and increasing the ability to detect contami-
nation of product lots containing very few contaminated
units.
4. Low-level contamination can be concentrated on the mem-
brane by ltering large volumes of product. This results
in faster reporting of test results since MF requires only
30 Chapter 1
7 days incubation (for most terminally sterilized prod-
ucts).
5. Organisms present in an oleaginous product can be sepa-
rated from the product during ltration and cultured in a
more desirable aqueous medium.
Conversely, the MF method presents two major disad-
vantages compared to the DT method:
1. There exists a higher probability of inadvertent contami-
nation in manual operations because of the need for
greater operator skill and better environmental control in
disassembling the ltration unit and removing, cutting,
and transferring the membrane. (Newer systems such as
the Steritest have eliminated this disadvantage.)
2. The method is unable in differentiate the extent of con-
tamination between units, if present, because all product
contents are combined, ltered through a single lter, and
cultured in single test tubes. Also, if accidental contamina-
tion has occurred, rather than this being detected in one
or more vessels of the DT method, it manifests itself in
the only container used per culture medium.
Interpretation of Results
If there is no visible evidence of microbial growth in a culture
medium test tube, after subjecting the sample and medium to
the correct procedures and conditions of the USP and EP ste-
rility test, it may be interpreted that the sample representing
the lot is without intrinsic contamination. Such interpretation
must be made by those having appropriate formal training in
microbiology and having knowledge of several basic areas in-
volved in quality control sterility tests:
1. Industrial sterilization methods and their limitations
2. Aseptic processing
3. Statistical concepts involved in sampling lots for represen-
tative articles
4. Environmental control procedures used in the test facility
If microbial growth is found or if the sterility test is judged to
be invalid because of inadequate environmental conditions,
the sterility test may be repeated. However, this introduces a
controversial and somewhat complicated subject.
STERILITY RETESTING
Sterility retests have been allowed by the USP since sterility
testing became a USP requirement (XI edition, 1936), but only
when the USP XX edition (1980) was published was there spe-
cic denitions of rst and second sterility retests. While ste-
rility retesting is allowed per the Code of Federal Regulations
(CFR), retesting without just cause is no longer allowed per
USP as of the eighth supplement, released in May 1998.
The FDA has repeatedly reafrmed that it supports the
USP position provided that industry shows due diligence in
their investigations of initial sterility test failures. Avallone
(24) wrote an FDA position paper in 1986 pointing out limita-
tions of the sterility test and, in fact, summarizing that a posi-
tive test result could indicate that the sample of product tested
was truly contaminated, while a negative test result does not
really mean that much, or even that the sample tested was
truly sterile. He describes the pharmaceutical industry as
having different levels of quality philosophy regarding steril-
ity tests. On one end there are manufacturers who recognize
the many limitations of aseptic processing and sterility test-
ing, so that if there is a sterility test failure, the batch is re-
jected. On the other end, there are manufacturers who will do
everything to justify the release of a product that fails an ini-
tial sterility test. Avallone goes on to discuss all the various
activities that a rm should consider when investigating an
initial sterility test failure and the decision-making process
that should be undertaken to release the lot of product. His
nal statement in this article is worth repeating: The man-
agement of a rm truly committed to quality has little if any
problem in the interpretation of sterility test results. Thus,
sterility retesting and investigation of initial sterility test fail-
32 Chapter 1
ures should be done with the highest degree of diligence and
responsibility on the part of high-level management of the par-
enteral industry.
FDA GUIDELINES ON STERILITY TESTING
The June 1987 FDA Guideline on Sterile Drug Products Pro-
duced by Aseptic Processing contains a fair amount of direc-
tion regarding conductance, evaluation, limitations, interpre-
tation, and retesting requirements of the USP sterility test.
The testing laboratory environment should employ facilities
and controls comparable to those used for the lling and clos-
ing operations (e.g., Class 100 air conditions for critical opera-
tions when a sterile product is exposed to the environment).
The limitations of the USP sterility test (as discussed on
pp. 5058) cause the FDA considerable concern with respect
to sampling plans and any positive test result that may occur.
In investigation of sterility test failures/positive test results,
the guidelines state: When persuasive evidence showing labo-
ratory error is absent, or when available evidence is inconclu-
sive, rms should err on the side of safety and batches should
be rejected as not conforming to sterility requirements. This
statement has caused much consternation among quality con-
trol (QC) groups in the pharmaceutical industry because as-
surance of sterility is so difcult to prove with absolute cer-
tainty.
Investigations of sterility test failures should consider ev-
ery single factor related to the manufacture of the product and
the testing of the product sample. Tables 1.101.12 show rep-
resentative lists of factors to be investigated by QC both in
the manufacturing areas and in the sterility test laboratory
to determine how a sterility test failure could have occurred.
Most of the time, there is no concrete conclusive evidence
pinpointing where the contamination occurred; thus, QC
must make a decision based on philosophical positions and ret-
rospective history of the manufacturing and sterility test
areas.
Table 1.10 Manufacturing Quality Control Checklist for
Investigating Sterility Test Failures (Aseptically Filled Products)
Item Code: Investigated by:
Lot No.: Disposition Date:
Line validation by media ll
Sterilization records of primary packaging components
Sterilization records of product contact or component contact
equipment
Environmental monitoring data
viable particle counts
nonviable particle counts
surface testing results
pressure differentials
log of room differentials
verication of laminar ow gauges in aseptic manipu-
lation zones
temperature
humidity
Bioburden of product-contact utilities
water
compressed gases (i.e., nitrogen, air)
clean steam
raw materials
HEPA lter certications for lters of the aseptic manipula-
tion zone
Sanitization logs
Sterilizing lter integrity test results for lters servicing the
following product contact utilities:
compressed gases
vent lters (vacuum)
Product sterilizing lter integrity test
Preventive maintenance and calibration records for critical pa-
rameter instruments of:
autoclaves
freeze dryers
Strunck tunnels
hot-air ovens
34 Chapter 1
Table 1.10 Continued
Corrective maintenance records and/or a physical inspection
of the equipment used to manufacture the lot in question, as
applicable (i.e., gaskets, joints, valves, piping, etc.)
Manufacturing ticket
Deviation reports, if any
Incoming sterility tests of purchased sterilized primary pack-
aging components
Vendor issues form Purchased Materials QC audits
Operator training records
personal broth test
gowning training
aseptic technique training
operator garment monitoring
Any other manufacturing-related parameter that might have
an impact on product sterility/integrity
Table 1.11 Manufacturing Quality Control Checklist
for Investigating Sterility Test Failures (Terminally
Sterilized Products)
Product sterilizing lter integrity tests
Preventive maintenance and calibration records for auto-
clave(s) used
Product terminal sterilization records
charts
loading sheets
Table 1.12 Quality Control Sterility Laboratory Checklist
for Investigating Sterility Test Failures
Test information
medium in which the growth was observed
ID of the isolate
number of lots or sections tested in the same test run
other tests showing contamination from the same test
run
analyst number of the individual performing the test
hood in which the test was performed
Sterilization records and BI results pertaining to the steriliza-
tion of the equipment and media used to perform the sterility
test of the lot in question
Environmental monitoring data
viable particle counts
analyst garment monitoring
Test controls
negative controls of the media used to perform the test of
the lot/section in question
manipulative negative control
equipment sterility check
media preparation and growth promotion records
Sanitization logs for
chemical sanitization of room
peracetic acid sanitization of room
HEPA lter certications for:
incoming air to room
laminar ow hoods
Analyst training records
analyst certication
gowning procedure
aseptic technique training
Lab retest data
by item, by line, by presentation
36 Chapter 1
The FDA aseptic guidelines indicate that persuasive evi-
dence of the origin of the contamination should be based on
the following:
1. The identication of the organism in the sterility test (ge-
netic typing may be useful or required)
2. The laboratorys record of tests over time
3. Monitoring of production area environments
4. Product presterilization bioburden
5. Production record review
6. Results of sterility retest
Thorough and complete review of all these data should
enable reviewers to determine the actual, or certainly proba-
ble, origin of the organism contaminating the sterility test
sample. In addition to the discussion below, Avallone (24) and
Lee (25) have written detailed articles on the evaluation and
investigation of initial sterility test failures.
Identication of the Organism in the Sterility Test
Not only the genus, but also the species of the isolated organ-
ism will provide invaluable information concerning the organ-
isms habitat and its potential resistance to the product for-
mulation and sterilization methods. If the organism is one
normally found on people, then the investigation can focus on
employee hygiene, washing and gowning techniques, and
aseptic techniques. Identication of the organism can be com-
pared to historical microbial databases for the manufacturing
and testing areas to assess probabilities of the origination of
the organism. Obviously, if the organism identied had been
isolated before in the production area, but never in the testing
area, then the production area would be implicated as the
source of the organism, and the test would be judged as a true
sterility test failure. Identication of the organism allows the
manufacturer to perform further testing to determine if the
organism is sensitive to the product formulation, particularly
if the product contains an antimicrobial preservative. If an or-
ganism that was isolated from a product that was terminally
sterilized and that had resistance to terminal sterilization is
proven to be below the microbial reduction produced by the
sterilization cycle, then it can reasonably be deduced that the
organism did not originate from the product. Knowledge of
whether the organism identied is an aerobe or anaerobe
would be important if the product were one that contained an-
tioxidants or was overlayed with nitrogen. For example, if the
organism were a strict anaerobe and the product was ushed
with nitrogen prior to sealing, then it must be strongly sus-
pected that the organism originated during the manufactur-
ing process and was protected by the nitrogenation of the
product.
The Laboratorys Record of Tests Over Time
The FDA nds it normal for a sterility testing laboratory to
have an initial positive sterility test failure of less than 0.5%
of all sterility tests. Therefore, if the laboratory shows a failure
rate higher than 0.5%, then problems must exist in the labora-
tory, production areas, or both. Trends of sterility test failures
should be noted, and rates of sterility test failure should be
grouped according to product type, container type, lling line,
and degree of manual manipulation. If a product is terminally
sterilized and the sterility test failure rate shows an upward
trend, then problems in the testing environment or personnel
can be suspected and action taken to eliminate the problem.
Conversely, upward trends in test failure of a product or line
of products manufactured aseptically can indicate production
problems that should be investigated. Not only monitoring of
failure rates as a function of time, but also monitoring of envi-
ronmental test data in both the production and testing areas
can provide important information for follow-up to correct po-
tential sources of contamination and keep the false-positive
failure rates extremely low.
Monitoring of Production Area Environments
Every manufacturing area should have a thorough and com-
plete record of environmental monitoring data obtained daily
as the area is being used. Trend analysis should be done on
38 Chapter 1
air-monitoring data (agar settle plates, viable air sampling,
particulate matter analyses) and surface monitoring (Rodac,
swab). Also, monitoring data of production personnel should
be closely scrutinized to ensure that personnel are practicing
good aseptic techniques.
Recent FDA inspections have resulted in 483 citations ei-
ther because the manufacturer either failed to have sufcient
environmental data or because such data were not used prop-
erly in assessing the acceptability of a lot that failed the steril-
ity test initially. For example, environmental data showed
contamination on the line used to ll a product that subse-
quently failed a sterility test, but such contaminants on the
line were different fromthat found in sterility test failure. The
rmrepeated the sterility test and released the lot on the basis
that the contaminant in the initial sterility test failure did not
come from the manufacturing area since it was different from
contaminants found in environmental testing of the area. The
FDA concluded that just because a contaminant found in a
sterility test failure was not specically identied in the man-
ufacturing environment, it does not preclude the contami-
nants presence in that manufacturing area. The rms envi-
ronmental monitoring data indicated the presence of a
problem, and the rm did not, according to the FDA, react re-
sponsibly toward this problem. Unequivocal proof must exist
to invalidate an initial sterility test positive, and the FDA will
heavily scrutinize any lot that is released that has failed a
sterility test. A thorough investigation with sound scientic
proof should be performed before repeating or releasing a
batch of material that has failed the sterility test.
Product Presterilization Bioburden
The bioburden of each lot of product should be known before
such product is sterilized either by terminal sterilization or by
aseptic ltration. Both methods have limits of sterilization. A
sterilizing grade lter with a nominal pore size of 0.22 m is
usually capable of retaining a bioburden of no more than 10
7
CFUper cm
2
as a function of lter surface area, while terminal
sterilization is typically capable of a 10
6
log reduction of bio-
burden. Trend analysis of product bioburden will determine
if an upward trend in bioburden might be occurring and if this
might be a contributor if sterility test failures also are showing
an upward trend over time.
Production Record Review
All records of producing a batch of product should be reviewed
(see Tables 1.101.12). Any one of these records showing an
aberrant result (e.g., high particulate counts, lack of proof that
a certain operation was done, a sterilization run that had prob-
lems, etc.) could help in determining the point at which this
aberrancy had an impact on product sterility.
Results fromall the above investigations should be collec-
tively reviewed by competent and experienced personnel and
reviewed by upper management, preferably a representation
from a variety of departmentsQC, production, technical ser-
vices, quality assurance (QA)before a nal decision is made
regarding the acceptance or rejection of a batch of product that
initially failed a sterility test.
It is rare that a determinant cause can be identied for
many sterility test failures, and it is an FDA expectation that
a specic cause be identied before a sterility test can be re-
peated. Some reasons to invalidate a sterility test may be
failed environmental monitoring in the sterility testing room
or isolator, a negative control that is positive after incubation,
cracked media containers, poor aseptic technique during the
sterility test, improper sanitization or sterilization of the ste-
rility testing area, defective stoppers and/or seals on sterility
samples, and so on.
The lot may only be released if a denitive reason is dis-
covered to invalidate the initial failure. Only then can the test
be repeated and the lot released. According to USP XXV, the
test may only be repeated if the positive sterility test result
can without a doubt be ascribed to faulty aseptic techniques
40 Chapter 1
or materials used in conducting the sterility testing proce-
dure.
STERILITY TESTING OF DIFFERENT STERILE PRODUCTS
The USP describes the sterility test procedures to be followed
for all types of sterile products; in addition, biologics, human
antibiotics, and veterinary biologics must comply with federal
regulations (9CFR113.26). Test procedures for the DT to test
media are given for the following ve types of products from
USP:
1. Nonlterable Liquids
2. Ointments and oils insoluble in isopropyl myristate
3. Solids to test media
4. Puried cotton, gauze, surgical dressings, sutures, and re-
lated articles
5. Sterilized devices
Test procedures for using the MF technique are specied
for the following nine types of products:
1. Liquids miscible with aqueous vehicles
2. Liquids immiscible with aqueous vehicles, less than 100
ml per container
3. Ointments and oils soluble in isopropyl myristate
4. Prelled syringes
5. Solids for injection other than antibiotics
6. Antibiotic solids for injection
7. Antibiotic solids, bulks, and blends
8. Sterile aerosol products
9. Devices with pathways labeled sterile
For a complete description of these test procedures, refer
to the appropriate section of the USP or EP. In the following
discussion, each test procedure is summarized and additional
information not found in the USP description is provided to
enhance the readers understanding and appreciation of the
procedure.
Direct Transfer
Nonlterable Liquids
Summary of procedure: Agitate the containers and aseptically
withdraw, from a sufcient number of units, the volumes
of mediumindicated. Mix each test specimen with the ap-
propriate medium, but do not aerate excessively. Proceed
to add each of the contents to TSB and FTM, respectively.
Commentary: The risk of inadvertent contamination is at its
greatest during this process. Strict aseptic technique
must be practiced to minimize this risk. Also, it must be
realized that a nite probability exists that the pipet or
syringe and needle may themselves not be sterile.
The USP cautions against excessive mixing of the
test sample and the medium. This is especially true for
FTM because of the need to preserve the efcacy of the
thioglycollate antioxidant, which maintains anaerobiasis
in the upper part of the vessel.
At least 14 days (for the USP test) or 21 days (for
the EP test) are required to ensure that any microbial
contaminants, if present, have been given sufcient time
to adapt to the FTM or TSB environment and to begin to
thrive and reproduce.
Ointments and Oils Insoluble in Isopropyl Myristate
Summary of procedure: 100 mg (or the entire contents if less
than 100 mg) from each of 10 containers are aseptically
transferred to a ask containing 200 ml of a sterile aque-
ous vehicle. Then, 20 ml of this mixture is mixed with 200
ml of FTMand incubated for not less than 14 days (for the
USP test) or 21 days (for the EP test). The entire process
described above is repeated with another 10 containers
using TSB as the medium.
42 Chapter 1
Commentary: Two key facets of this procedure are (a) em-
ploying strict aseptic technique in the two transfer pro-
cesses for each medium, and (b) choosing the correct dis-
persing agent in the aqueous vehicle that both adequately
disperses the oil or ointment homogeneously in the vehi-
cle and, in the concentration used, has no antimicrobial
capacity in and of itself. The most commonly used dis-
persing agents are surface-active agents, such as Polysor-
bate 80 and Triton X-100, dissolved in water. Some feel,
however, that Triton X-100 exerts an antimicrobial effect.
Solids to Test Media
Summary of procedure: Transfer a quantity of product in the
form of a dry solid (or prepare a suspension of the product
by adding sterile diluent to the immediate container) in
the quantity directed in the USP. Transfer the material
so obtained to 200 ml of FTM and mix. Repeat the above
procedure for TSB. The incubation time is again not less
than 14 days (for the USP test) or 21 days (for the EP
test).
Commentary: Most sterility testing facilities prefer reconsti-
tuting the dry sterile solid with sterile water for injection.
The chance of accidental contamination is greatly en-
hanced because of the extra manipulations involved in
reconstituting and then withdrawing the uid sample for
transfer. The adherence to strict aseptic technique cannot
be overemphasized.
Puried Cotton, Gauze, Surgical Dressings, Sutures,
and Related Articles
Summary of procedure: From each package of cotton, rolled
gauze bandage, or large surgical dressings being tested,
aseptically remove two or more portions of 100 to 500 mg
each from the innermost part of the sample. From indi-
vidually packaged, single-use materials, aseptically re-
move the entire article. Immerse the portions or article
in each medium, and incubate the test for not less than
14 days (for the USP test) or 21 days (for the EP test).
Commentary: If the entire article is too large to be transferred
intact to culture media, a suitable portion or the inner-
most part of the article is tested and assumed to be repre-
sentative of the entire article. The rationale for selecting
the innermost part is the fact that this part is the most
difcult area for steam or gas to penetrate during termi-
nal sterilization. Strict aseptic technique must be fol-
lowed when performing the above test to avoid inadver-
tent contamination of the test medium. Since most
devices are packaged in paper or Tyvek, which is perme-
able to gas sterilants, the use of isolators to test devices is
often not feasible because of the gas sterilization system
associated with them.
Sterilized Devices
Summary of procedure: Articles can be immersed intact or dis-
assembled. To ensure that device pathways are also in
contact with the media, immerse the appropriate number
of units per medium in a volume of medium sufcient to
immerse the device completely and incubate the test.
For catheters where the inside lumen and outside
are required to be sterile, either cut them into pieces such
that the medium is in contact with the entire lumen or
ll the lumen with medium and then immerse the intact
unit.
For extremely large devices, immerse those portions
of the device that are to come into contact with the patient
in a volume of medium sufcient to achieve complete im-
mersion of those portions.
In all cases, DT items must incubate for a total of
not less than 14 days (for the USP test) or 21 days (for
the EP test).
Commentary: If the entire article is too large to be transferred
intact to culture medium, a suitable portion or the inner-
most part of the article is cut out and assumed to be repre-
44 Chapter 1
sentative of the entire article. The rationale for selecting
the innermost part is the fact that this part is the most
difcult area for steam or gas to penetrate during termi-
nal sterilization. Strict aseptic technique must be fol-
lowed when performing the above test to avoid inadver-
tent contamination of the test medium. Again, since most
devices are packaged in paper or Tyvek, which is perme-
able to gas sterilants, the use of isolators to test devices is
often not feasible because of the gas sterilization system
associated with them.
Membrane Filtration
Liquids Miscible with Aqueous Vehicles
Summary of procedure: At least 20 or 40 containers of product
are used. Sufcient volumes required for both media are
transferred aseptically into the membrane lter funnel
or withdrawn from the container using the Millipore
Steritest system. Vacuum or pressure is applied, and the
solution is ltered. The membrane is removed aseptically
and cut in half; one half is placed in 100 ml of FTM, while
the other half is placed in 100 ml of TSB. In the case of
the Steritest system, the lter is contained within the
chamber that is incubated. Incubation is carried out for
not less than 7 days (for products that are terminally ster-
ilized by moist heat) or 14 days (for products that are re-
quired to meet EP standards or are not terminally steril-
ized by moist heat).
For (a) LVP solutions, (b) 50 ml to less than 100 ml
for intravenous use, or (c) between 100 ml and 500 ml,
the entire contents of 10 containers are aseptically trans-
ferred and ltered through each of two lter assemblies,
or if only one lter assembly is used, or 20 containers
are emptied or withdrawn from the container using the
Millipore Steritest system. For LVP solutions with vol-
umes greater than 500 ml, at least 500 ml are transferred
from each of 10 containers through each of two lter as-
semblies or from each of 20 containers if one lter assem-
bly is used. Then, the membrane is removed using sterile
forceps, cut in half with sterile scissors, and the halves
aseptically added to 100 ml of FTM and TSB, respec-
tively.
With the high usage frequency of total parenteral
nutrition solutions in hospital practice, many LVPs are
nowavailable containing high concentrations of dextrose.
These and other highly viscous solutions are ltered
through several lter assemblies since one assembly will
not permit the passage of the entire contents of a viscous
solution. However, the total volumes and number of con-
tainers per medium remain the same as required for non-
viscous solutions. Half of the total number of membranes
used are incubated in each medium.
Commentary: While the MF method offers distinct advantages
over the DT method, the risk of extraneous contamina-
tion is greatly increased because of the manipulations ad-
ditional to those employed in conducting the sterility test
by DT. Thus, extreme precautions must be followed in all
the techniques involved in the MF method. Negative con-
trols are especially recommended with the above method-
ology. Use of apparatuses such as Steritest and isolation
systems have greatly reduced the number of manipula-
tions required by the operator in using the MF technique.
These systems thus have greatly helped minimize adven-
titious contamination and strengthened the advantages
of the MF technique.
In transferring the container contents into the mem-
brane lter funnel, great care must be used to avoid
squirting solution directly onto the lter. Also, since this
method is used to sterility test small-volume multidose
parenterals containing antimicrobial preservatives, the
membrane must be rinsed three times with USP Diluting
Fluid A (100 ml) to ensure that the entire solute content
has been washed through the membrane. The MF method
is an excellent technique for the sterility testing of LVP
46 Chapter 1
solutions because low levels of contaminants in these di-
lute solutions are concentrated together on the surface of
one or two lters. If only the direct transfer method were
available, even a representative sample of LVP added to
culture media would contain an insufcient number of
microbial cells to harvest under the best of incubation
conditions.
Liquids Immiscible with Aqueous Vehicles (Less than 100 ml per
Container)
Summary of procedure: The required volume from 20 contain-
ers is transferred aseptically directly into one or two sep-
arate membrane lter funnels. After ltration via vac-
uum, the membrane is cut in half using the aseptic
procedure already described and is incubated in 100 ml
each of FTM and TSB. For immiscible liquids of high vis-
cosity, aseptic addition of Diluting Fluid D is required to
increase the ow rate. If the liquid has antimicrobial ac-
tivity or contains an antimicrobial preservative, the lter
is washed three times with 100 ml of the diluting uid.
Products containing lecithin, however, must use Diluting
Fluid D containing the surface-active agent Polysorbate
80 to enable the dispersion of the oily substance.
Commentary: Examples of products tested by this procedure
are progesterone, testosterone propionate, and dromosta-
nolone propionate, in which the solvent is sesame oil or
peanut oil.
Ointments and Oils Soluble in Isopropyl Myristate
Summary of procedure: Dissolve not less than 100 mg from
each of 20 units (or 40 units if the contents are not suf-
cient for each medium) in 100 ml of isopropyl myristate
that previously has been rendered sterile by ltration
through a sterilizing membrane lter. Warm the sterile
solvent, and if necessary the test material, to a maximum
of 44C just prior to use. Swirl the ask to dissolve the
ointment or oil, taking care to expose a large surface of
the material to the solvent. Filter this solution promptly
following dissolution, keeping the lter membranes cov-
ered with the solution throughout the ltration for maxi-
mum efciency of the lter. Wash the membranes with
two 200-ml portions of Fluid D, then wash with 100 ml
of Fluid A. The media used in the test should contain 0.1%
Polysorbate 80.
If the substance under test contains petrolatum, use
Fluid K, moistening the membranes with about 200 ml
of the uid before beginning the ltration. Keep the mem-
branes covered with the prepared solution throughout the
ltration operation for maximum efciency of the lter.
Following ltration of the specimen, wash the mem-
branes with three 100-ml volumes of Fluid K. Incubate
the test membranes as mentioned in the previous step.
Commentary: The use of Polysorbate 80 in the nal test me-
dium is required to facilitate total dissolution of the oil
or ointment being tested so that any organisms present in
the material will not be isolated fromthe growth medium.
Isopropyl myristate was found to be a satisfactory solvent
for dissolving petrolatum-based ointments without ad-
versely affecting contaminants (27). Filter-sterilized iso-
propyl myristate is less toxic to microorganisms than
heat-sterilized isopropyl myristate (28,29). Another sol-
vent system that has been reported to aid in the sterility
testing of parenteral fat emulsions is dimethyl-sulfoxide
(DMSO) (30).
Prelled Syringes
Summary of procedure: For prelled syringes without
attached sterile needles, expel the contents of each sy-
ringe into one or two separate membrane lter funnels
or into separate pooling vessels prior to transfer. If a sep-
arate sterile needle is attached, directly expel the syringe
contents as indicated above and proceed as directed for
liquids miscible with aqueous vehicles.
48 Chapter 1
Fig. 1.6 Example of a large plastic syringe dispenser unit.
Commentary: The frequency of use of prelled disposable sy-
ringes has increased signicantly in recent years. In
many instances, syringes are prelled in hospital phar-
macies. Therefore, hospital pharmacists must be trained
in performing sterility tests and maintaining proper
aseptic techniques in performing the tests. Aseptic tech-
nique is also especially important in cases when a needle
must be attached later to the prelled syringe.
Solids for Injection Other than Antibiotics
Summary of procedure: Constitute the test articles as directed
on the label and proceed as directed for liquids miscible
with aqueous vehicles or liquids immiscible with aqueous
vehicles, whichever applies. Excess diluent may be added
to aid in the constitution of the test article.
Commentary: Some solids for injection may not be soluble in
a solvent suitable for the sterility test; consequently, the
DT test must be employed.
Antibiotic Solids for Injection
Summary of procedure: For pharmacy bulk packages, less
than 5 g, from each of 20 containers, aseptically transfer
about 300 mg of solids into a sterile 500-ml conical ask,
dissolve in about 200 ml of Fluid A, and mix; or, consti-
tute, as directed in the labeling, each of 20 containers and
transfer a quantity of liquid or suspension, equivalent to
about 300 mg of solids, into a sterile 500-ml conical ask,
dissolve in about 200 ml of Fluid A, and mix. Proceed as
directed for liquids miscible with aqueous vehicles or liq-
uids immiscible with aqueous vehicles, whichever ap-
plies.
For pharmacy bulk packages 5 g or larger, from each
of 6 containers, aseptically transfer about 1 g of solids
into a sterile 500-ml conical ask, dissolve in about 200
ml of Fluid A, and mix; or, constitute, as directed in the
labeling, each of 6 containers and transfer a quantity of
liquid or suspension equivalent to about 1 g of solids into
a sterile 500-ml conical ask, dissolve in about 200 ml of
Fluid A, and mix. Proceed as directed for liquids miscible
with aqueous vehicles or liquids immiscible with aqueous
vehicles, whichever applies.
Commentary: It is important that all antimicrobial properties
of the antibiotic be removed or inactivated when per-
forming the test for sterility. Successful bacteriostasis/
fungistasis validation must be performed prior to the ste-
rility test. See Table 1.15 for a list of inactivators com-
monly used with antibiotics.
Antibiotic Solids, Bulks and Blends
Summary of procedure: Aseptically remove a sufcient quan-
tity of solids from the appropriate amount of containers,
mix to obtain a composite equivalent to about 6 g of solids,
and transfer to a sterile 500-ml conical ask; dissolve in
about 200 ml of Fluid A and mix. Proceed as directed for
liquids miscible with aqueous vehicles.
50 Chapter 1
Commentary: Again, it is important that all antimicrobial
properties of the antibiotic be removed or inactivated
when performing the test for sterility. Successful
bacteriostasis/fungistasis validation must be performed
prior to the sterility test. See Table 1.15 for a list of inacti-
vators commonly used with antibiotics.
Sterile Aerosol Products
Summary of procedure: For uid products in pressurized aero-
sol form, freeze the containers in an alcoholdry ice mix-
ture at least at 20C for about 1 hour. If feasible, allow
the propellant to escape or puncture the container before
aseptically adding the contents to a sterile pooling vessel.
Add 100 ml of Fluid D to the pooling vessel and mix gen-
tly. Proceed as directed for liquids miscible with aqueous
vehicles or liquids immiscible with aqueous vehicles,
whichever applies.
Commentary: The Millipore Corporation offers a Steritest sys-
tem for use with aerosol products. This method offers
many advantages over the above method, including that
the Steritest system tests not only the liquid or active
drug product, but also the propellant; there is no need
to puncture the aerosol container, which could result in
inadvertent contamination; and there is no need to freeze
the container. The FDA recently proposed a rule to re-
quire all solutions for nebulization and/or inhalation to
be sterile due to recalls of contaminated inhalation solu-
tions (103).
Devices with Pathways Labeled Sterile
Summary of procedure: Aseptically pass nor less than 10 path-
way volumes of Fluid D through each device tested. Col-
lect the uids in an appropriate sterile vessel and proceed
as directed for liquids miscible with aqueous vehicles or
liquids immiscible with aqueous vehicles, whichever ap-
plies.
In the case of sterile, empty syringes, draw sterile
diluent into the barrel through the sterile needle, if
attached, or through a sterile needle attached for the pur-
pose of the test and expel the contents into a sterile pool-
ing vessel, then proceed as directed above.
Commentary: Many device manufacturers rely on a paramet-
ric release by which the sterility test is not performed in
lieu of biological indicator data. Biological indicators are
included in each sterilization load of the device to be pro-
cessed and must be negative to meet the requirement of
sterility.
STERILITY TESTING OF ANTIBIOTICS AND PROTEINS
Antibiotics and Antimicrobial-Containing Products
The MF method for sterility testing was developed as a solu-
tion to the antimicrobial properties of antibiotics. After ltra-
tion of the antibiotic or antimicrobial product, rinsing of the
membrane is essential to remove any residual antibiotic. USP
Fluid A or Fluid D are rinsing uids of choice. For penicillins,
penicillinase is added to facilitate antibiotic inactivation. The
amount of penicillinase added is determined experimentally.
One important consideration in the sterility testing of an-
tibiotics and penicillin is the issue of containment. Testing of
penicillin must be in a laboratory separate from other test
facilities, and cephalosporin sterility testing facilities should
also be dedicated facilities.
Proteins
Gee et al. (32) described a modication of MF system for l-
tration of large volumes of viscous protein solutions (e.g., 25%
w/v normal serumalbumin). This modication operates under
intermittent positive pressure through a set of membrane l-
ter canisters.
Insulin zinc forms precipitates in sterility test media
(both SCD and FTM) (6). Ascorbic acid at 1% in 0.1% peptone
52 Chapter 1
(w/v) dissolves protamine zinc insulin and insulin zinc in no
more than 1 minute without harming organisms.
CONTROL IN STERILITY TESTING
Sterility testing provides an estimate of the probable extent of
contamination of a lot of articles. Since it is only an estimate, it
must be based on sound scientic principles. Such principles
primarily involve the successful incorporation of controls
within each test. Sterility testing is, however, only one compo-
nent of control of sterility (sterility assurance in manufac-
ture). In the broadest sense, control starts with the environ-
mental, personnel, and sterilization conditions implemented
during the manufacture of the sterile product. Control of the
quality of the environment under which the sterility test is
performed is of extreme importance. The training and experi-
ence of personnel conducting the sterility test must also be
controlled with regard to their understanding, use, and atti-
tude toward strict aseptic technique. These types of controls
in manufacture are discussed in a separate section. The types
of control of sterility testing to be discussed in this section in-
clude the following: (a) positive control of the culture media
(that is, the testing of the growth-promoting quality of each
lot of media); (b) negative control of the culture media (that
is, testing the sterility of the media); (c) control of the product
itself (that is, obtaining knowledge about the bacteriostatic
and/or fungistatic activity of the product prior to its being sub-
jected to a sterility test); and (d) specic controls when using
the MF technique.
Positive Controls
The absence of growth in sterility test samples at the comple-
tion of the test indicates that the product is sterile insofar as
assumptions and limitations of the test are considered, that
is, it meets the requirements of the test. However, this conclu-
sion can be made only with the assurance that growth would
have occurred during the sterility test period had microorgan-
isms actually been present. The USP and EP growth promo-
tion tests are designed to serve as a positive control for each
lot of sterility test media. Each lot is inoculated with 10 to 100
of the microorganisms listed in Tables 1.13 and 1.14. Growth
of these microorganisms must occur in the appropriate me-
dium within 7 days of incubation. The evidence of growth in
duplicate test containers compared with the same lot of me-
dium containing no microbial inoculum qualies the test me-
dium to be used for sterility test purposes. The USP allows
for the growth promotion test to be the positive control run
simultaneously with the actual sterility test with the under-
standing that the test becomes invalid if the medium does not
support the growth of the inoculated microorganisms. How-
ever, if tested media are stored, additional tests are prescribed
for particular storage conditions.
Negative Controls
Negative controls consist of containers of culture media with-
out addition of product sample or microbial challenge. The
purpose of negative control samples is to verify the sterility of
the medium before, during, and after the incubation period of
the sterility test. If microbial growth is detected with a nega-
tive control, the medium was not sterilized properly, contami-
nation was introduced accidentally during the test procedure,
or there exists an inefciency in the container or packaging
system. If such microbial growth in a negative control occurs
and in the absence of evidence from the environmental moni-
tor, equipment, or personnel of accidental contamination, it
becomes a clear indication for retesting the product.
Bacteriostatic and Fungistatic Testing
If a sterility test is negative (no growth), there must be the
assurance that growth was not inhibited by the antimicrobial
properties of the product itself. The USP provides a procedure
for determining the level of bacteriostatic and fungistatic ac-
tivity of a product or material prior to its being tested for ste-
5
4
C
h
a
p
t
e
r
1
Table 1.13 Test Microorganisms Required by the USP for Use in the Growth Promotion
and Bacteriostasis/Fungistasis Test Used in Sterility Testing
a
Incubation
Medium Test microorganism
b
temperature (C) Condition
Fluid thioglycollate Staphylococcus aureus (ATCC 6538)
c
32.5 2.5 Aerobic
Pseudomonas aeruginosa (ATCC 9027)
d
32.5 2.5 Aerobic
Clostridium sporogenes (ATCC 11437)
e
32.5 2.5 Aerobic
Alternative thioglycollate
f
Clostridium sporogenes (ATCC 11437) 32.5 2.5 Anaerobic
Soybean-casein digest Bacillus subtilis (ATCC 6633) 22.5 2.5 Aerobic
Candida albicans (ATCC 10231) 22.5 2.5 Aerobic
Aspergillus niger (ATCC 16404) 22.5 2.5 Aerobic
All organisms are required to show visible growth within not more than 7 days of the original test.
a
ATCC cultures represent reference species, and their use for compendial test is predicated on them not being sub-
jected to procedures that may alter their properties. Such procedures include indenite numbers of subcultures with
no standardization of conditions. For this reason, the USP has proposed that seed lot culture techniques may be
used and that the viable microorganisms used be no more than ve passages removed from the reference species.
b
Available from the American Type Culture Collection, 10801 University Boulevard, Manassas, VA 20110-2209.
c
An alternative to Staphylococcus aureus is Bacillus subtilis (ATCC 6633).
d
An alternative microorganism is Micrococcus luteus (ATCC 9341).
e
An alternative to Clostridium sporogenes, when a nonspore-forming microorganism is desired, is Bacteroides vulga-
tus (ATCC 8482).
f
Used for sterility test of devices that have tubes with small lumens.
S
t
e
r
i
l
i
t
y
T
e
s
t
i
n
g
5
5
Table 1.14 Test Microorganisms Required by the EP for Use in the Growth Promotion
and Bacteriostasis/Fungistasis Test Used in Sterility Testing
a
Incubation
Medium Test microorganism
b
temperature (C) Condition
Fluid thioglycollate Staphylococcus aureus (ATCC 6538)
c
32.5 2.5 Aerobic
Pseudomonas aeruginosa (ATCC 9027)
d
32.5 2.5 Aerobic
Clostridium sporogenes (ATCC 19404)
e
32.5 2.5 Aerobic
Soybean-casein digest Bacillus subtilis (ATCC 6633) 32.5 2.5 Aerobic
Candida albicans (ATCC 10231) 22.5 2.5 Aerobic
Aspergillus niger (ATCC 16404) 22.5 2.5 Aerobic
Note that there is no alternative uid thioglycollate medium as in USP XXV.
a
ATCC cultures represent reference species, and their use for compendial test is predicated on them not being sub-
jected to procedures that may alter their properties. Such procedures include indenite numbers of subcultures with
no standardization of conditions. For this reason, the USP has proposed that seed lot culture techniques may be
used and that the viable microorganisms used be no more than ve passages removed from the reference species.
b
Available from the American Type Culture Collection, 10801 University Boulevard, Manassas, VA 20110-2209.
c
An alternative to Staphylococcus aureus is Bacillus subtilis (ATCC 6633).
d
An alternative microorganism is Micrococcus luteus (ATCC 9341).
e
All bacteria are required to demonstrate visible turbidity within 3 days of incubation, and all fungi are required
to demonstrate visible turbidity within 5 days of incubation.
56 Chapter 1
rility by the DT or MF test. Basically, the procedure calls for
adding product to containers of culture media in volumes cor-
responding to those that would be used for testing the product
containing 10 to 100 of the microorganisms listed in Table 1.13
and comparing with medium-inoculum controls without the
product. If the material possesses bacteriostatic or fungistatic
activity, then the product-media will show decreased or no mi-
crobial activity compared to control culture media. If this is
the case, then procedures must take place for the proper inac-
tivation of these bacteriostatic/fungistatic properties. Either
a suitable sterile inactivating agent must be found or the ma-
terial and medium must be adequately diluted to overcome
the static effects. If at all possible, the MF test should be ap-
plied for those materials found to be bacteriostatic or fungi-
static. When MF is used, similar comparisons are made of in-
cubated lters through which product and suitable diluting
uid have been passed, each containing the same added micro-
organisms.
Specic inactivating or diluting methods used for a few
drugs or drug products known to be bacteriostatic or fungi-
static are listed in Table 1.15.
Controls for Membrane Filtration Techniques
The MF test relies on the ability to produce sterile equipment
and to have aseptic conditions under which to conduct the test.
Three basic control procedures are recommended in separate
experiments:
1. The membrane lters are challenged after their steriliza-
tion cycle for their ability to retain microorganisms.
2. The exposure times for agar settling plates used to moni-
tor the environment are validated.
3. The cleaning procedures used to remove bacteriostatic
and/or bactericidal residues from equipment following the
MF test must be validated. This is especially important
for the equipment involved in the sterility testing of anti-
biotics.
Table 1.15 Inactivation of Bacteriostatic/Fungistatic Agents
in Sterile Products Tested by the Direct Transfer Sterility Test
Agent Method of inactivation
Mercurials
Phenylmercuric nitrate 10 ml FTM
(1: 50,000 conc.)
Merthiolate (1: 10,000 conc.) 10 ml FTM or 12% sodium thio-
sulfate
Phenol Adsorb on 0.1% Darco or 0.03%
ferric chloride or dilute 0.5%
phenol in 50 ml culture
medium
Benzalkonium chloride Lecithin and polysorbate 80
Sulfonamides p-Aminobenzoic acid
Penicillin Penicillinase
Cephalosporins Cephalosporinase
Streptomycin Cysteine HCl 2% in acid medium
Cresol Dilute 0.35% in 60 ml culture
medium
Chlorobutanol Dilute 0.5% in 40 ml culture
medium
Barbiturates Dilute to 0.2% in culture medium
with a pH of about 7.0
Aminoglycosides Acetyl-coenzyme A
a
a
A. S. Breeze and A. M. Simpson, An improved method using acetyl-coen-
zyme Aregeneration for the enzymatic inactivation of aminoglycosides prior
to sterility testing, J. Appl. Bacteriol., 53, 277 (1982).
VALIDATION OF THE STERILITY TEST
For every product that is tested for sterility, the sterility test
method must be validated for that product. What this means,
simply, is that prospective validation studies must be per-
formed to collect data to prove that the sterility test can detect
microbiological contamination in the product. Validation of
the sterility test for a particular product involves adding small
but known concentrations (100 CFU) of various microorgan-
isms to the nal rinse and then demonstrating recovery of the
58 Chapter 1
organisms using the sterility test methodology. Table 1.13 pro-
vides the test organisms required by USPXXV. Table 1.14 pro-
vides the test organisms required by the EP. While the EP
and USP chapters on sterility testing are now considered to
be harmonized, if the practitioner desires to test a product
for sterility release and that product is required to meet EP
and USP requirements, there are several key points to con-
sider.
1. Organisms from tables 1.13 and 1.14 will have to be used
in the bacteriostasis and fungistasis test, and Bacillus
subtilis must be tested in both FTM and TSB. Upon incu-
bation of the challenge containers, all bacteria must show
visible growth within 3 days of the test, and all fungi must
show visible growth within 5 days of the test.
2. Even if the product is terminally sterilized, the nal steril-
ity test must incubate for 14 days (if the MF technique is
used) to satisfy the EP requirement.
3. ATCC 19404 must be used for the Clostridium sporogenes
challenge to satisfy the EP requirement, while ATCC
11437 must be used for the C. sporogenes to satisfy the
USP requirement when performing the bacteriostasis and
fungistasis test.
4. When the DT test is employed, the initial transfer test is
required to incubate for 14 days for EP, but only 7 days
for the USP test.
LIMITATIONS OF THE USP/NF REFEREE STERILITY TEST
The USP referee sterility test suffers from at least three limi-
tations: (a) the invariant uncertainty that the small sample
used in the test reliably represents the whole lot, (b) the inabil-
ity of the culture media and incubation conditions to promote
the growth of any and all potential microbial contaminants,
and (c) the unavoidable problem of occasional accidental con-
tamination of the sterility test samples. Ernst et al. (33) be-
lieve that impeccable control of three phases of the steriliza-
tion and sterility testing of parenteral products will alleviate
many of the problems of sterility testing: (a) knowledge and
understanding of the sterilization process, (b) avoidance of un-
favorable environmental conditions during manufacture and
testing, and (c) education of personnel in the procedures of ste-
rility testing.
The Problem of Sampling and Statistical Representation
The probability of accepting lots having a given percentage
contamination is related to the sterility test sample size rather
than to batch size (34). For example, if a batch is 0.1% contam-
inated (1 nonsterile unit in 1000 units) and 10 units are sam-
pled for a sterility test, the probability of nding 1 of those 10
samples to be the 1 contaminated unit in 1000 is not signi-
cantly different if the batch size were 1000, 2000, or 5000).
Increasing the sample size from 10 to 20 to 50 units per batch,
however, affects the probability of accepting the batch as ster-
ile to a more signicant degree than does the increase in batch
size, assuming that the increase in batch size does not increase
the level of contamination. This phenomenon is depicted in
Table 1.16. The probability rate does not change as the batch
size is increased, but does change as the sample size is in-
creased. Of course, a key factor is that the contamination rate
remains at 0.1% as the batch size increases. This, in reality,
may not be true, especially for aseptically lled products.
Hence, as the contamination rate increases with batch size, the
probability of acceptance decreases for the same sample size.
Table 1.16 Probability of Accepting a Batch as Sterile
Assuming the Contamination Rate to be Constant at 0.1%
Batch size
Sterility test
sample size 1000 2000 5000
10 0.99 0.99 0.99
20 0.98 0.98 0.98
50 0.95 0.95 0.95
60 Chapter 1
The relationship of probability of accepting loss of varying
degrees of contamination to sample size is given in Table 1.17
(35). Three details may be learned assuming the data in Table
1.17 are real:
1. As the sample size is increased, the probability of ac-
cepting the lot as sterile is decreased.
2. At low levels of contamination, (e.g., 0.1%), the odds of
ever nding that 1 contaminated sample in 1000 units are
so small that one must face the fact that lots are going to
be passed as sterile but somewhere, at some time, some
patient is going to receive that nonsterile sample (even at
a contamination rate of 1% with 20 sterility test samples,
it must be realized that such a lot will be passed as sterile
82% of the time).
3. Realistically, a batch must be grossly contaminated for the
sterility test to detect it. This fact was concluded at a 1963
conference on sterility testing in London (36), at which ex-
perts in sterility testing recognized that the lowest con-
tamination rates that can be detected with 95%condence
are 28% with a sample size of 10, 15% with a sample size
of 20, and 7% with a sample size of 40 units.
A sample size of 20 units is shown in Table 1.18. As an
example, if it is assumed that only 1 unit in a batch of 100,000
units is contaminated (0.001%), the probability that the 1 con-
taminated unit is among the 20 sterility test samples taken
at random is 0.0002, or 2 times in 1 million sterility tests. Ta-
ble 1.19 presents an example of why dependence on sampling
and sterility is, in fact, a futile attempt to prove the sterility
of a lot.
If there is contamination, an acceptable level of accep-
tance is 1 in 100 lots tested, the probability that 0.1%of the lot
is contaminated is achieved with 40 samples, but with lower
contamination levels, many more samples would be required.
For example, sample requirements for 0.01% contamination
would be 450 samples; for 0.001% contamination, it is 4500
samples; and for 0.0001% contamination, it is 45,000 samples.
S
t
e
r
i
l
i
t
y
T
e
s
t
i
n
g
6
1
Table 1.17 Relationship of Probabilities of Accepting Lots of Varying Assumed Degrees
of Contamination to Sample Size
Probability of accepting the lot (true percentage contamination of lot)
Number of
samples tested (n) 0.1 1 5 10 15 20
10 0.99 0.91 0.60 0.35 0.20 0.11
20 0.98 0.82 0.36 0.12 0.04 0.01
30 0.95 0.61 0.08 0.01
100 0.91 0.37 0.01
300 0.74 0.05
500 0.61 0.01
62 Chapter 1
Table 1.18 Probability of Finding at Least One
Nonsterile Unit in a Sample Size of 20 Subjected
to a Sterility Test
Assumed percentage Probability of
nonsterile units nding at least
in the lot 1 nonsterile unit
0.10 0.01980
0.05 0.00995
0.02 0.00399
0.01 0.00199
0.005 0.00100
0.002 0.00040
0.001 0.00020
This illustrates the futility of attempting to determine steril-
ity levels (when low) by sterility tests alone.
Take the following example: If sterility tests have been
done, using one medium and 20 samples on each occasion, and
only 2 inoculated tubes showed growth, the proportion con-
taminated may be 2/200. However, if when a positive result
Table 1.19 Futility of Depending on Sampling and Sterility
Tests for Sterility Assurance of a Lot
Probability of nding all negatives in samples of
different sizes for various levels of contamination
n p .1 p .01 p .001
10 0.35 0.90 0.99
20 0.12 0.82 0.98
40 0.01 0.69 0.96
160 0.20 0.85
640 0.53
If the proportion of contaminated units in the lot is p, then the proportion
of noncontaminated units in that lot is 1 p. Let that be designated q. Then,
the probability of nding noncontamination (i.e., acceptance) of that lot with
taking n samples for testing is (q)
n
.
was obtained, the tests were repeated with another 20 sam-
ples each time, with negative results, the proportion contami-
nated may be 2/240, that is, 0.0083, and the proportion not
contaminated 0.9917. If that lot were contaminated to the de-
termined level, to reduce the probability of acceptance to 1 in
100 would require about 550 samples to be taken. Not only
is such a number not feasible, the probability of adventitious
contamination in sterility tests (ranging from 0.2% to 3% of
tests, see the section Problem of Accidental Contamina-
tion) makes even that possibility likely to yield an unreliable
result.
A mathematical equation for calculating the probability
P of releasing lots at different levels of contamination was de-
veloped by Armitage (37):
P e
mv
where m is the number of microorganisms per milliliter, and
v is the volume in milliliters of the test sample. For example,
if 10 microorganisms are present per 100 ml and the test sam-
ple is 100 ml (20 containers 5 ml per container), the proba-
bility of releasing the lot of this contaminated product is
0.0000454. Like the presented data in the preceding tables,
this equation shows that relatively small sample sizes and/or
low contamination levels result in lots being judged to meet
the sterility test requirements when, in fact, a nite number
of articles in the lot are nonsterile. Thus, claims for low proba-
bility levels of nonsterility cannot realistically be proven by
the random sampling procedure of the USP sterility test, and
sterility assurance must be achieved by appropriate control
measures in manufacture (see the section, Support Tech-
niques and Procedures for Sterility Assurance). In fact, with
low levels of nonsterile units in a lot, any reasonable sampling
plan would not provide realistic results (see Table 1.19). This
does not even consider the nite probability of inadvertent
contamination entering the product during the sterility test
procedures.
64 Chapter 1
Problem of Supporting the Growth
of Microbial Contaminants
No single medium will support the growth of all microbial
forms (that is, bacteria, molds, fungi, and yeasts). FTM will
not recover very low levels of some aerobic spore formers such
as Bacillus subtilis (14). Friedl (38) reported that TSB gave
more efcient recovery of small numbers of B. subtilis and C.
sporogenes spores than in FTM. TSB, being strictly an aerobic
medium, will not support the growth of the genus Clostridia.
On the other hand, while FTM effectively supports the growth
of various strains of Clostridia, it has been reported that so-
dium thioglycollate is toxic to Clostridia, and this antioxidant
should be replaced by cysteine hydrochloride (39).
TSB is incubated at 2025C to permit adequate growth
of facultative organisms such as enterobacteria (Escherichia
coli, Salmonella, Shigella, Proteus, Serratia marcescens, and
Flavobacterium) and many yeasts. FTM is incubated at 30
35C to detect mesophilic bacteria. These sterility media,
therefore, are not incubated at temperatures conducive to the
growth of psychophiles (predominantly pseudomads) and
thermophiles (predominantly bacilli). According to Bruch (40),
TSB and FTM do not contain the necessary nutritional ingre-
dients to support the growth of obligate halophiles, osmo-
philes, or autotrophs.
Problem of Accidental Contamination
Growth that occurs in sterility test media must be ascertained
to have originated from the test sample and not from the cul-
ture media or from an external source during the execution of
the test. Such a determination can be made only to a limited
extent. The use of negative controls eliminates one source of
contamination, that resulting from nonsterile culture media.
Thus, a positive sterility test result is concluded to be true (the
test sample is contaminated) unless it can be shown to be false
(contamination was accidently introduced during the test pro-
cedure). The problem of false positives is widespread and can-
not be completely eliminated.
The percentage of false-positive sterility tests is reported
in the range from 0.2% to 3% (1,2,41). In a poll conducted by
one of us, of 10 pharmaceutical companies involved in sterile
product sterility testing, the range of inadvertent contamina-
tion found during sterility testing was 0.1% to 5%. The most
common types of microbial contaminants found in false-posi-
tive sterility test samples are listed in Table 1.20.
False-positive sterility tests result also from contami-
nants located in the environment (air and surfaces), on people
conducting the test (hands, breath, hair, clothing, etc.), or on
the equipment used in conducting the test (nonsterile mem-
brane ller assemblies, scissors, forceps, lters, etc.). Contam-
ination being accidently introduced by the environment can
be reduced signicantly by performing a monitored environ-
mental sterility test in a laminar air ow (LAF) workbench
(see the section, Laminar Air Flow). For example, Parisi and
Table 1.20 Examples of Microorganisms Found in Aseptically
Filled Parenteral Products (Found in Antibiotic and Nonantibiotic
Products)
Microorganism type Source Examples
Gram positive cocci Human contamination Staphylococcus
Micrococcus
Streptococcus
Gram negative bacilli Water Pseudomonas
Coliforms (also GMB) Fecal contamination Escherichia
Enterobacter
Citrobacter
Gram positive bacilli Dirt, dust Bacillus
Clostridium
Corynebacterium
Gram negative cocci Rare pathogens Neisseria
Molds Air, dust Penicillium
Aspergillus
Yeast Air, dirt Candida
Rhodotorula
Saccharomyces
66 Chapter 1
Borick (42) found that the percentage of false positives during
sterility testing fell from 1.61% when done in conventional
sterile rooms to 0.63%when done in an LAF workbench. These
same authors also reported that 2361 colonies were recovered
on 765 agar settling plates located in the conventional sterile
room, while only 75 colonies were recovered on 299 agar set-
ting plates located on an LAF workbench. Thus, while LAF
workbenches do not completely eliminate the incidence of con-
tamination, they do signicantly reduce the potential problem
provided the results from the settling plates are used to indi-
cate corrective actions.
The single largest contributor of accidental contamina-
tion in sterility test samples is the person or people performing
the test. Personnel-induced accidental contamination pri-
marily results from a lack of strict adherence to good aseptic
technique. Good aseptic technique involves many consider-
ations, including apparel, eye-hand coordination, concentra-
tion, and the desire to be as careful as possible. An excellent
resource for training on aseptic techniques is the chapter by
Luna (43).
Accidental or adventitious contamination is one of the
greatest problems interfering in the interpretation of sterility
test results in hospital pharmacies (44). Bernick et al. (45)
suggested that contaminated intravenous admixtures are not
contaminated during the admixture process, but rather are
contaminated frommicroorganisms introduced during the ste-
rility testing procedure. Such admixture processing should be
carried out in appropriate hospital pharmacy facilities and not
in the patient care areas. Sterility testing should be an essen-
tial component in the monitoring of intravenous solutions and
admixtures in hospital pharmacy practice (4651).* Several
* This has been more recently substantiated by the Draft Guidelines on Quality As-
surance for Pharmacy-Prepared Sterile Products (Am. J. Hosp. Pharm., 49, 407417,
1992), which propose that pharmacists establish quality assurance procedures, in-
cluding sterility tests, for pharmacies that are involved in the preparation of sterile
products.
methods have been suggested for evaluating sterility of intra-
venous admixtures (5256). However, the problem of adventi-
tious contamination and the limitations resulting from this
problemthat affect the interpretation of the sterility test must
be recognized. The National Coordinating Committee for
Large Volume Parenterals (NCCLVP) strongly recommended
that suitable education programs in hospitals and colleges
be developed to educate and train personnel involved in the
preparation and administration of sterile medication (49).
NCCLVP also recommends developing procedures for in-use
testing of LVPs suspected of contamination.
ISOLATION CHAMBERS AND ROBOTIC STERILITY
TEST UNITS
As previously discussed, false-positive sterility tests occur be-
cause of inadvertent contamination of the sample in the steril-
ity test laboratory. Such contaminations are of a nite proba-
bility as long as human manipulation is involved. Concerns
over such unreliabilities of the sterility test have given rise to
new technologies designed to remove as much as possible the
human element involved in sterility testing.
dArbeloff et al. (57) have described four robotic sterility
test systems: (a) HoffmannLa Roche rst documented ro-
botic system in the pharmaceutical industry (58); (b) Far-
mitalia; (c) Takeda; and (d) Precision Robots, Incorporated,
Autotest 1000. Undoubtedly, there are other commercially
available or company-built systems being used today. Each ro-
botic system is designed to eliminate the tasks that contribute
to or cause the inadvertent contamination problems causing
false-positive sterility tests. Each of the robotic systems de-
scribed by dArbeloff et al. (57) was successful in eliminating
adventitious contamination. For example, the Roche system
had not had a conrmed false-positive test in over 4 years of
operation.
Major disadvantages of the robotic sterility test systems
are their expense of installation and maintenance, slower
68 Chapter 1
speed in conducting sterility tests, and increased complexity
of setting up, using, and maintaining the system.
Validation of robotic sterility test systems involves at
least ve elements:
1. Validating the current manual procedure, which should
be a procedure used over many years and having a known
and low rate of adventitious contamination.
2. Validating all the robotic operations, both hardware and
software.
3. Validating the laminar air ow patterns in the system.
4. Validating the particle levels in and around the test areas.
5. Validating the disinfection of the system. Peracetic acid
should not be used as a disinfectant in robotic systems (it
can be used in isolation chambers) because of it causes
corrosive problems. Acceptable disinfectants for robotic
systems include Ampl, Sporicidin, or 3% hydrogen per-
oxide.
Finally, dArbeloff et al. (57) recommend that the robotic
system be challenged by introducing contamination, (e.g.,
worst-case operator contamination in introducing vials into
the system) to show that the design of the system, its opti-
mized laminar air ow, and the removal of any further human
intervention indeed remove seeded microbial contamination
on product containers.
While the LaCalhene isolation chamber was the most
widely used sterility test chamber system in the parenteral
industry in the 1980s to mid-1990s, the use of hard-walled iso-
lators has become the most recent trend in sterility testing
(see Fig. 1.7) (102). The LaCalhene module (Figure 1.8) is
made of polyvinyl chloride supported externally by a frame-
work of stainless steel rods. The barriers can be accessed by
the operator through either glove sleeves or half-suits. Materi-
als can be introduced into or removed from these barriers
through a double door transport port sterile transfer door.
Room air enters and exits through a 0.3-m high-efciency
Fig. 1.7 Example of a Schaeffer Engineering hard-walled isolation
barrier system for sterility testing (courtesy of Baxter Pharmaceuti-
cal Solutions LLC, Bloomington, Indiana).
particulate air (HEPA) lter. LaCalhene offers many different
types of isolation chambers in terms of design and function.
All sterility test operationsproduct container surface decon-
tamination, sterility test manipulations, and incubation of
samplesoccur within the barrier system.
One of the major aspects of the isolation chamber is the
sterilization and its validation of all surfaces within the cham-
ber and product containers and other items brought into the
chamber. The original method of surface sterilization was the
use of peracetic acid as a spray. Davenport (59) described
the use of peracetic acid as a sterilizer in LaCalhene sterility
testing chambers and the methods used to validate that the
sterilant is effective in destroying the biological indicator
(Bacillus circulans UC9951).
70 Chapter 1
Fig. 1.8 Example of a LaCalhene isolation barrier system for ste-
rility testing (courtesy of Eli Lilly and Co., Indianapolis, Indiana).
The most commonly used method of surface sterilization
today is VPHP (vapor phase hydrogen peroxide) (102). The
VHP 1000
manufactured by the Steris Corporation (formerly

AMSCO) (see Fig. 1.9) is widely used in the pharmaceutical
industry in conjunction with barrier isolation systems. VPHP
is less corrosive to metals such as stainless steel than is perox-
yacetic acid.
The advent of isolation chambers and robotic sterility test
systems has challenged the long-held level of acceptability of
false positives. The current level of 0.5%rate of false positives,
both in aseptic manufacture and sterility testing, may no
longer be acceptable as it is becoming more plausible to have
a much lower rate of contamination. Cooper (60) points out
that the industry may be facing a two-tier situation in which
Fig. 1.9 The VHP 1000
unit used for surface sterilization within

isolators (courtesy of Steris, Erie, Pennsylvania).
companies that have implemented isolation chambers or ste-
rility test robotics will be expected to maintain lower levels of
false positives than companies that still perform sterility tests
under the conventional laminar owlaboratory method. It will
be very interesting to see how this situation will be addressed
by the FDA and other regulatory bodies in the years ahead.
Validation of Barrier Isolation and Associated
Sterilization Systems
Like any other process in the pharmaceutical industry, barrier
isolation used for sterility testing must be shown to reproduci-
bly deliver the desired result. Because of their complexity,
there are several parameters to consider in the design and val-
idation of isolation systems. USP chapter 1208 provides
guidance for the design and validation of isolator systems for
use in sterility testing. The guidelines in USP 1208 are
72 Chapter 1
summarized below, as are common practices in the validation
of sterility testing isolators. For a complete description of
1208, consult the most current USP.
The steps and considerations that are essential to the iso-
lator validation and design are outlined below:
Design
By design, an isolator which is used for sterility testing should
be equipped with lters capable of microbial retention. HEPA
lters are required, but ULPA (ultra-low penetration air) l-
ters may be substituted. While the isolator is at rest, it must
meet the particulate requirement for a class 100 area as de-
scribed in U.S. Federal Standard 209E (71). There is no partic-
ulate requirement while the unit is in operation during a ste-
rility test, and there is no requirement for air velocity or air
exchange rate. The isolator should be leak-proof, but it may
exchange air with the surrounding environment. While direct
openings with to the surrounding environment should be
avoided, air overpressure can be employed to maintain sterile
conditions within the isolator. Air overpressure should also be
employed to help avoid ingress of nonsterile air in the event
of an unexpected leak.
Location of the Isolator
The isolator does not need to be installed in a classied clean
room, but the surrounding room should be limited to essential
staff. Environmental monitoring of the surrounding room is
not required.
The surrounding room should have sufcient tempera-
ture and humidity control to maintain operator safety and
comfort, to allow for proper operation of the associated steril-
izer (the air should exhaust to an outside source for safety rea-
sons) unit, and to allow for proper operation of the isolator.
The temperature within the roomshould be as uniformas pos-
sible to avoid the formation of condensation within the iso-
lator.
Installation Qualication
The installation qualication (IQ) should include a detailed
description of all of the mechanical aspects of the system, such
as dimensions, internal conguration, serial numbers of the
equipment, blueprints, purchase orders, electrical supply,
specications, exhaust, vacuumsupply, and equipment manu-
als. All documentation should be reviewed for accuracy. The
following pieces of documentation are recommended:
Equipment: The equipment is listed with critical design speci-
cations. The IQ should verify that the appropriate de-
sign specication was received and that all equipment
was installed per the manufacturers requirements.
Construction materials: The critical components of the system
are checked for compliance with the design specication
and for compatibility with the method of sterilization.
Instruments: System instruments are listed with their cali-
bration records.
Utility specications: All utilities that are required for opera-
tion as dened in the operating manuals and diagrams
are veried. Any connections between electrical and/or
exhaust systems are inspected and veried to conform to
specications.
Filter certication: HEPA lters are tested and certied, and
copies of the certications are included.
Computer software: All computer software is listed with
name, size, and version number. Any master copies
should be properly labeled and stored should the need for
a backup arise.
Operational Qualication
The operational qualication (OQ) step veries that the isola-
tor system operates within conformance to functional aspects.
Operational performance check: All alerts and alarms should
be tripped and proper functioning veried.
Isolator integrity check: The integrity of the isolator should be
74 Chapter 1
veried to be free from leaks. The leak test is important
to preclude contamination and for operator safety. The
overpressure set point should be established and shown
to be maintained during operation.
Sterilization cycle verication: Verication of relevant tem-
perature and/or humidity control during the sterilization
cycle should be demonstrated. Humidity may be espe-
cially important depending on the type of sterilizing gas
that is used. The concentration and distribution of the
sterilizing gas should be measured using chemical indica-
tors. After sterilization, verication that the sterilizing
gas has been removed (by aeration) to an acceptable level
should be demonstrated by quantitative methods.
Sterilization cycle development: On completion of the OQ, cy-
cle development parameters are established to achieve
sterilization of the isolator unit. Sterilization should be
demonstrated by use of the bioburden approach, through
the use of biological indicators (BIs) of a known concen-
tration, or by the half-cycle approach.
Note that Bacillus stearothermophilus is an appropriate
choice as a BI in this application, as it is more resistant to
VPHP than most environmental isolates (104).
Performance Qualication
The performance qualication (PQ) veries that the systems
are functioning in compliance within its operational require-
ments. On completion of the PQ phase, sterilization efcacy
should be established. For sterilization validation, the interior
surfaces of the isolator, articles within the isolator, equipment
in the isolator, and sterility test articles should be rendered
sterile after processing. BI kills of 10
3
to 10
6
are commonly
used in the pharmaceutical industry in association with steril-
ity testing isolators.
Since most pharmaceutical companies produce many dif-
ferent package presentations, the minimum load (an empty
chamber) and maximum load (the maximum number of test
articles) approach is helpful when performing the PQ of the
isolator.
False-Negative Evaluation
Because certain materials are adversely affected or absorb
sterilizing agents, it is important to demonstrate that the
method of sterilization would not destroy any microbial life
since the point of a sterility test is to detect lowlevels of organ-
ism in the nished product.
Various containers, media, and rinsing uids should be
inoculated with low levels of organisms (100 CFU) and sub-
jected to the sterilization cycle (suggested organisms are listed
in Table 1.13). After sterilization, the level of inoculum should
be veried to prove that organisms are not destroyed during
the cycle. In the event that organisms are adversely affected,
a new container closure should be selected before use with the
isolator system.
SUPPORT TECHNIQUES AND PROCEDURES
FOR STERILITY ASSURANCE
Because sterility assurance is based on a probability function,
sterility can never be proven unless the entire contents of a lot
are subjected to a sterility test. Even this is not theoretically
possible because of the need to use at least two different media
for the test. In addition, as already discussed, the sterility test
itself has certain limitations. Therefore, product sterility can-
not be tested with absolute assurance that every container of
sterile product is sterile. However, assurance of product steril-
ity can be achieved with a high degree of probability by the
employment of and adherence to various procedures, of which
sterility testing is only an adjunct. These include (a) sterilizer
and sterilization method validation using physical and biologi-
cal indicators; (b) impeccable control of the environmental con-
ditions under which the parenteral product is manufactured,
particularly when aseptic processing is performed; and (c)
thorough training of personnel of the strict aseptic techniques
76 Chapter 1
required for performing the sterility test. Any sterility test
should be done in an environment equal to or more controlled
than that used for aseptic processing. An important part of
long-range sterility assurance is adequate documentation of
the validation, monitoring, personnel training/use of aseptic
technique, and batch manufacturing procedures used.
Sterilizer and Sterilization Method Validation
The assurance of parenteral product sterility primarily de-
pends on the process used to sterilize the product. The greater
the control of the process, the greater the assurance of steril-
ity. Sterilization process control involves knowledge and man-
agement of process variables such as temperature, pressure,
concentration, humidity, load conguration, and lter integ-
rity and of product variables such as solution composition and
viscosity, packaging specications, and microbial content.
Four basic methods are employed to sterilize parenteral
products.
1. Heat, both wet (steam) and dry
2. Gas, primarily ethylene oxide
3. Radiation, primarily cobalt 60, gamma irradiation, and
electron beam
4. Filtration through bacterial retentive membrane bers
The mechanics and engineering of each of these processes
must be understood and properly controlled for the process to
provide additional assurance of product sterility.
Simmons (61) has elaborated on the engineering aspects
of validating steam, dry heat, and ethylene oxide sterilizers.
Filter integrity testing has been adequately described by Reti
and Leahy (62).
Once the sterilizing system itself has been qualied (i.e.,
for capability to achieve sterilization), then the process of ster-
ilization can be validated. Validation of the process involves
both physical and biological methodology. Physical methods
include temperature measurement, gas concentration or irra-
diation dose monitoring, and the use of mathematical expres-
sions such as the F value equation (63). Biological methods
involve the employment of biological indicators to evaluate the
ability of the sterilization process to destroy or eliminate an
inordinately high concentration of known resistant microor-
ganisms under conditions identical to those found in the steril-
ization of the actual parenteral product. They also are used to
monitor a validated sterilization cycle.
Biological Indicators
Greater condence in sterility assurance has arisen because
of the increased acceptance and employment of BIs during the
development of the sterilization cycle or system (64). If the
sterilization process is shown with a high degree of probability
to destroy, say, 10
6
spores of known resistance to the process,
then a batch of parenteral product exposed to that same pro-
cess will result in a sterile product. This may be roughly con-
rmed by the sterility test. BIs are microorganisms, usually
spore forms, known to be as resistant to destruction by a given
sterilization process as any microbial form known. BIs are
used to verify the effectiveness of a sterilization process be-
cause if the process can destroy the BI of known concentration,
it is assumed that the process will also destroy all other micro-
bial contaminants potentially present in the product. Of
course, this assumption is controversial, and many experts
question how far one can really depend upon it.
Microorganisms recognized by the USP as BIs for the var-
ious sterilization processes are given in Table 1.21. However,
one is not restricted from employing other types of microor-
ganisms as BIs if they better serve the needs of the particular
process. Several species of Bacillus spore are known to be
more resistant than the strain of Bacillus subtilis var. niger
(ATCC 9372), the USP biological indicator for monitoring eth-
ylene oxide sterilization (66). B. pumilus (ATCC 27142) has
demonstrated the same degree of resistance to ethylene oxide
as B. subtilis var. niger (66).
Vegetative cells, rather than bacterial spores, are em-
ployed in testing and validating ltration sensitization. Bre-
7
8
C
h
a
p
t
e
r
1
Table 1.21 Performance Characteristics of Biological Indicators on Paper Strips
Sterilization Approximate Survival time Kill time
Culture process D value spore content (not less than) (not more than)
Bacillus stearo- Saturated steam 1.31.9 10
6
3.9 min 19 min
thermophilus at 121 0.5C min
spores (ATCC VPHP
7953 or 12980)
Bacillus subtilis Ethylene oxide at 2.65.8 10
6
7.8 min 58 min
var. niger 54 2 and rel- min
(ATCC 9372) ative humidity
60 10%; 600
30 mg/liter
Dry heat at 160 1.31.9 10
6
3.9 min 19 min
5C min
Bacillus pumilus Ionizing radiation
(ATCC 27142)
Wet preparations 0.160.24 10
6
0.6 Mrad 2 Mrad
Mrad
Dry preparations 0.120.18 10
6
0.45 Mrad 1.5 Mrad
Mrad
General D 20% 10
6
3 D 10 D
requirement
vundimonas diminuta (ATCC 19146), a vegetative organism
selected for its small size (approximately 0.3 m), is the organ-
ism of choice for evaluating the retention ability of 0.2-m
sterilizing membrane lters. Recent controversy has arisen
over the ability of a 0.2-m membrane to retain all organisms;
therefore, the use of 0.1-m lters or redundant 0.2-m ltra-
tion has become more common in the pharmaceutical in-
dustry.
The USP provides a general description of BIs, which are
available either as liquid suspensions or as dried preparations
on carriers such as paper strips, glass, plastic beads, ampules,
or stainless steel strips. BIs used as a spore suspension should
be added to representative units or to units similar to those of
the lot to be sterilized. The BI must demonstrate a challenge
of the sterilization process that exceeds the challenge of the
natural bioburden. BIs must be properly standardized so that
the BI units in the lot all exhibit the same degree of resistance
to the sterilization process when used in the same manner,
even if varying at different times within the dating period of the
BI. The BI inoculummust be prepared under the supervision of
trained microbiologists in order to maintain and standardize
BI cultures of known purity, identity, and resistance. Every
commercially prepared BI product must be labeled according
to the relevant USP general notices on labeling, as well as with
its spore content and performance characteristics, such as deci-
mal reduction time (D value) under given sterilization parame-
ters, directions for use, and recommendations for disposal.
Several interesting review articles have been written on
the principles and applications of BIs. Borick and Borick (3)
were the rst to write a lengthy discussion on the use of BIs
versus the use of regular sterility test samples. Bruch (40) pre-
sented a strong case for using BIs as a means of evaluating
the probability of sterility of products sterilized by methods
other than saturated steam under pressure. Myers and Chrai
(67) reviewed the biology of microbial resistance and applica-
tion of bioindicators in designing and monitoring sterilization
cycles.
80 Chapter 1
Caputo and Mascoli (68) suggested a four-step process in
the design of a BI system for validating the efcacy of a steril-
ization cycle. In the rst step, the microorganism to be used
as the BI is selected and propagation procedures are developed
to ensure the consistent production of a homogeneous popula-
tion of BI with the desired resistance to the sterilization pro-
cess. Second, the D value (the time required to reduce the
microbial population by 90% or through one log cycle) is deter-
mined for the selected BI. Factors that must be considered and
that affect the D value of a particular BI were discussed by
Pug and Odlaug (69). The third step in the design of a BI
system is the actual evaluation of the sterilization process in
destroying the BI employing a full load of product. Process pa-
rameters (temperature, gas concentration, humidity, radia-
tion dose) are established during this step. Finally, a determi-
nation is made either of the amount of (log cycle) reduction
required for the desired degree of probability or of the level of
microorganisms to be used as a BI to validate the sterilization
process, qualify the sterilization vessel, and, subsequently,
monitor the sterilization process (70).
Environmental Control
Sterility tests should be performed in a test area that conforms
to Class 100 conditions as described by Federal Standard
Number 209E (71). Class 100 conditions mean that no more
than 100 particles per cubic foot of size 0.5 m or greater, as
measured by electronic particle counters, shall be found in the
measured area. A comparison of the classes of air cleanliness
is provided in Table 1.22. However, these classes refer to levels
of particulate matter, not viable microorganisms. The 1987
Guideline on Sterile Drug Products Produced by Aseptic Pro-
cessing recommends no more than 25 organisms per 10 cubic
feet of air for a Class 100,000 area, and no more than 1 organ-
ism per 10 cubic feet for a Class 100 area. No recommendation
for a Class 10,000 microbial specication is made in the docu-
ment, though a value of organisms per 10 cubic feet is
common.
Great strides have been made in recent years to help en-
sure that Class 100 conditions are met and that adequate mi-
crobial monitoring is effected in a sterility testing facility.
Probably the greatest advancement was the discovery by
Whiteld (72) in 1961 of the concept of laminar air ow.
Laminar Air Flow
Phillips and Miller (73) have succinctly described the concept
of laminar air ow (LAF). The employment of LAF cabinets,
workbenches, and rooms in the proper execution of the steril-
ity test and other aseptic operations is essential. The air emit-
ted from LAF equipment is claimed to be 99.97% free from
microbial contamination. This level is based on the removal of
dioctylphthalate particles 0.3 m and larger. Thus, although a
theoretical 0.03% contamination level exists when using LAF
equipment, the air within the conned area of the workbench
or cabinet is considered to be sterile.
LAF equipment can deliver clean air in a vertical, hori-
zontal, or curvilinear direction. The principles of vertical and
horizontal air ow are shown in Figs. 1.10 and 1.11, respec-
tively. Room air is taken into the equipment and passes
through a prelter, which removes large-size air contami-
nants. A blower then forces the preltered air through a sec-
ond lter system in the LAF unit called a High Efciency Par-
ticulate Air (HEPA) lter (Fig. 1.11). Air passing through the
HEPA lter not only is 99.97% particle free, but also moves
with uniform velocity along parallel ow lines. Proper aseptic
procedures are to be practiced while working at the laminar
ow workbench during sterility.
Quality control procedures must be adopted to evaluate
and monitor the quality of the LAF hood environment. This
includes monitoring with particles and microorganisms. Since
LAF hoods are supposed to provide Class 100 air, they should
be certied that this standard is met. Certication is done im-
mediately after installation of new HEPA lters and at peri-
8
2
C
h
a
p
t
e
r
1
Table 1.22 Guidelines for Air Cleanliness Classes
Type of facility Class 100 Class 10,000 Class 100,000
Laminar air ow
Vertical ow room, Entire work area meets Entire area meets re- Entire area meets re-
vertical ow cur- requirements at nor- quirements quirements
tain units, vertical mal working height
ow bench locations
Crossow room, tunnel First work locations Entire work area meets Entire area meets re-
room, wall-to-oor meet requirements requirements if parti- quirements
room, crossow cle generation, work
bench locations, and person-
nel are reasonably
controlled.
Nonlaminar air ow
Conventional clean Will not meet require- Can be upgraded to Will meet requirements
room ments under opera- meet requirements with strict observa-
tion conditions by placing laminar tion of rules govern-
air ow devices ing personnel, opera-
within the room and tions, garmenting,
continously ltering and janitorial proce-
the recirculating air. dures
Personnel and opera-
tion restrictions and
janitorial mainte-
nance are also re-
quired.
S
t
e
r
i
l
i
t
y
T
e
s
t
i
n
g
8
3
Rooms containing Will meet requirements Entire area meets re- Entire area meets re-
computer systems with personnel requirements. quirements
striction and janito-
rial maintenance
Maximum number of particles Maximum number of particles
per cubic foot that are per cubic foot that are
0.5 m and larger Class 5.0 m and larger
100 100 0
1000 1000 7
10,000 10,000 65
20,000 20,000 130
100,000 100,000 700
1,000,000 1,000,000 6,500
Source: Liberty Industries, East Berlin, Connecticut.
84 Chapter 1
Fig. 1.10 Vertical laminar air ow bench (courtesy of Liberty In-
dustries, East Berlin, Connecticut).
Fig. 1.11 Horizontal laminar air ow bench (courtesy of Liberty
Industries, East Berlin, Connecticut).
odic intervals, usually every 6 months. The velocity of HEPA-
ltered air is measured using an air velometer. Air velocity in
all parts of the lter should be 90 20 feet per minute. Air
quality is evaluated using particle counters, microbial air
samplers, and agar settling plates. The efciency of the HEPA
lter in removing particulate and microbial contamination is
evaluated by employing the dioctylphthalate (DOP) test. This
is a universally acceptable challenge test for HEPA lters.
DOP is a volatile liquid that, under pressure, converts to a
vapor or smoke having a size range of 0.28 to 0.4 m. The DOP
smoke is introduced at the supply plenum. Aphotometer probe
then scans the entire HEPA lter surface. Any leaks in the
lter will permit the DOP smoke to escape, and this will be
detected by the photometer. Several references are available
that describe the testing of laminar ow equipment (7577).
Phillips and Runkle (78) have published a comprehensive re-
view of the biomedical applications of LAF.
For most sterility testing operations, horizontal LAF
units appear to be superior to vertical ow hoods because the
air movement is less likely to wash organisms from the opera-
tors hands or equipment into the sterility test media (73).
However, the operator must be specically trained on how to
utilize the air owproperly. HEPA-ltered air will not sterilize
the surface of a contaminated object. It will only maintain ste-
rility or cleanliness of an already sterile or clean object. All
surfaces of the LAF hood except the HEPA lter itself must
be thoroughly disinfected before placing any item inside the
hood. All materials must be disinfected prior to introducing
such materials onto the surface of the LAF workbench. For
example, all glassware, containers, and other articles with sur-
faces that are nonsterile must be wiped thoroughly with a dis-
infectant solution before placing these items in the LAF unit.
Sterile materials enclosed in protective packaging, such as
plastic bags contained in polyethylene outer pouches, dispos-
able syringes, sterile scissors and forceps wrapped in alumi-
num foil, wrapped membrane lter units, and the like may be
introduced into the LAF unit by removing the outer protective
86 Chapter 1
Fig. 1.12 Direction of air ow in a laminar air ow work station
(courtesy of Liberty Industries, East Berlin, Connecticut).
package at the edge of the workbench before placing the sterile
itemon the workbench surface. Understanding of the LAF pat-
tern is very important to avoid turbulence and blockage of
HEPA ltered air reaching the critical work site (Fig. 1.12).
Design and Maintenance of Aseptic Areas
The USP states that the principal sources of contamination
are the air and water in the aseptic processing area and the
personnel, materials, and equipment involved in the pro-
cessing. Avis (79) has described considerations that must be
met in the design, construction, and implementation of a ster-
ile products facility. An encyclopedia nowexists for the design,
operation, and all other aspects of clean rooms, white rooms,
and sterile rooms (80). All areas must be designed and con-
structed to permit adequate cleaning, efcient operation, and
comfort of personnel. Process ow must follow a plan by which
product and personnel move to increasingly clean environ-
ments. An example of a process ow diagram is found in Fig.
1.13 (79).
Fig. 1.13 Diagram of ow of materials through a production area
(from Ref. 79.)
Ceilings, walls, and oors in the aseptic processing area
must be sealed for ease and thoroughness in washing as well
as treatment with disinfecting products. All counters, cabi-
nets, and sinks should be constructed of stainless steel. All
equipment, service lines and facilities, and other essential
room xtures should be constructed in such a manner to per-
mit ease of cleaning and disinfecting and to prevent the accu-
mulation of dust and dirt.
Several engineering features of a well-designed aseptic
area are listed in the USP. However, newer proposals empha-
size the principles involved rather than describe in detail the
features of the facility. The following are a number of features,
not all of which are necessarily applicable to any particular
facility:
1. Hoods, cabinets, and other enclosures to serve as a pri-
mary barrier around the process, while the aseptic room
itself serves as a secondary barrier
2. Maintenance of differential positive air pressures to pre-
vent inward leakage of air
3. Effective ltration of air supplied to the primary and sec-
ondary barriers
4. Provision of air locks and/or air showers at the entrances
88 Chapter 1
to rooms, a gowning room, and adequate space garment
storage for personnel
5. An effective intercommunication system and a suitable
room arrangement to minimize trafc
An appropriate program should be established initially
to qualify aseptic areas and equipment and routinely to moni-
tor the integrity of the measures. The least controlled and po-
tentially greatest source of contamination originates from the
people working in the aseptic processing area and conducting
the sterility test procedures. Training of personnel to mini-
mize potential contamination arising from people is discussed
in the section, Personnel Training.
Methods of Evaluating the Environment
A number of proven, quantitative viable and nonviable micro-
bial counting methods are available for evaluating and moni-
toring the environment in which sterility tests are conducted.
These methods can be further divided into air- and surface-
sampling methods.
Air-Sampling Methods
Slit-Air Sampler. This is a device that collects viable air-
borne microbial and particulate contamination (see Fig. 1.14).
A 150-mm sterile agar plate containing a layer of sterile agar,
usually trypticase soy agar, is placed on a circular plate in the
slit-air device, and the cover containing the slit is secured
above the agar plate. The speed of the rotation of the plate
and the volume of air sampled can be adjusted to record the
desired rate and degree of contamination of the air environ-
ment. The split-air sampler is one of the most widely used
monitoring methods for sterile manufacturing and quality
control environments.
Liquid Impinger. This device works by using a vacuum
source to suck in air at a high velocity through an isotonic
impingement uid, then passing the uid through a mem-
brane lter by vacuum and incubating the lter on an agar
plate. While liquid impingement thoroughly collects viable mi-
Fig. 1.14 Slit-to-agar biological air sampler. (Courtesy of New
Brunswick Scientic Co., Inc., Edison, New Jersey.)
crobial contamination within a given cubic foot of air, it suffers
two primary disadvantages. One disadvantage is the fact that
microbial counts may be underestimated because the high ve-
locity of impingement kills many organisms on impact with
the agar surface. The other disadvantage is the problem of air
locks occurring at the lter surface as the impingement and
diluting uids are being vacuum ltered.
Electronic Air Particle Counters. These instruments count
all particles in the environment and cannot differentiate be-
tween viable and nonviable particles. These counters are espe-
cially useful in determining the number of particle counts per
cubic foot to classify the cleanliness of a particular room or
area.
Settling Plates. These represent the simplest means of
evaluating the microbiological quality of air. A 100-mm petri
dish containing trypticase soy agar or other suitable medium
90 Chapter 1
Fig. 1.15 Agar settling plate.
is placed in a sampling location with the lid removed as shown
in Fig. 1.15. The period of sampling is controlledusually 2
to 4 hoursbefore the lid is placed over the medium and the
plate incubated. Colonies are counted, and many different lo-
cations within the sterility testing and manufacturing areas
can be controlled and compared for microbial contamination.
The major disadvantage of this method is that the volume of
air sampled and represented on the agar plate is unknown.
Centrifugal Air Sampler. This device (see Fig. 1.16) is an-
other method used to determine airborne contamination. Air-
borne microbes approximately 16 inches above the sterile
drum housing are drawn toward the impeller blades. Then,
owing to the applied centrifugal force, 40144178 rpm, the mi-
crobial particles are impacted at high velocity onto the agar
surface of the agar strip wound around the impeller blades.
Fig. 1.16 Biotest RCS centrifugal air sampler.
After incubation, colonies on the strips are counted; the re-
sults reported as colony-forming units per unit volume of air.
The air capacity of the sampler per minute is 40 liters. In a
comparative study of airborne microbial recovery rates (81),
the RCS centrifugal air sampler was found to be signicantly
more efcient than the slit sampler and the liquid impinger.
The centrifugal air sampler samples a greater area (1.2 cubic
feet) versus 0.5 cubic feet sampled by the slit sampler (82).
92 Chapter 1
Fig. 1.17 SAS
air sampler (courtesy of Bioscience International,

Rockville, Maryland.)
Sieve Impaction Air Sampler. A relatively new develop-
ment in air sampling is a class of samplers known as the sieve
impaction air sampler (see Fig. 1.17). Air is drawn through
the sieve toward a cartridge holding nutrient medium. Any
organism present in the air sample will be deposited on the
agar surface of the cartridge. After the sample is collected, the
cartridge is removed and incubated. After incubation, colonies
are counted; the results are reported as colony-forming units
per volume of air. These units can typically sample a relatively
large volume (up to 1000 liters) in as little as 8 minutes.
Surface-Sampling Methods
Rodac Plates. These are specially built petri plates in
which sterile culture media, usually trypticase soy agar con-
taining Polysorbate 80 and lecithin, is poured onto a baseplate
until a convex surface extends above the rim of the baseplate
(see Fig. 1.18). Once the molten agar has solidied, the agar
surface can be gently pressed against a selected surface, for
example, the surface of an LAFworkbench. The lid is replaced,
and the plate is incubated for the required length of time at
a controlled temperature. Surface contamination can be quan-
tied by counting the colonies after incubation. The presence
Fig. 1.18 Convex surface of Rodac agar plate.
of Polysorbate 80 and lecithin serves two purposes: to aid in
the complete contact and removal of microbes from the sam-
pled surface and to permit cleaning of the sampled area with
water and/or a disinfectant solution.
Swab-Rinse Test. This is a simple surface sampling
method employing sterile cotton swab tips to sample locations
that are unwieldy for Rodac plates or difcult to reach. The
swabs are then placed into tubes of culture media or, for micro-
bial quantication, are mixed with sterile water and then a
sample of the water is placed on a solid agar plate.
Many of these environmental testing procedures and a
suggested program for determining microbiological burden
and action levels for both nonsterile and sterile environments
can be found in an article by Dell (83). Tables 1.23 and 1.24
are reproduced from Dells article. These guidelines can be
useful in establishing a program design for environmental
monitoring specic to the history, conditions, and needs of any
particular manufacturing and sterility testing facility.
9
4
C
h
a
p
t
e
r
1
Table 1.23 Example of an Environmental Testing Program for Monitoring a Sterile
Production Facility
Membrane Most probable Pour Settle
Element ltration number plate Rodac Swab plate Slit-air Other
a
Frequency
b
Walls and oors
Sterility test area X X M
Controlled areas X X M
Critical areas X X M
HEPA-ltered air
Controlled areas X X DOP M
Critical areas X X DOP B
Sterility test area X X DOP B
Components X X X MLT H, B
In-process bulk X X X MLT H, B
Production equipment
Tanks X
Hoses and lines X DM, VF M
Filling equipment VF
Compressed air and gas X ORG H
Potable water X X X LF D
Deionized and distilled X X X PT D
water
HEPA lter X X V, DOP M
Finishing supplies X X H
Source: Ref. 83.
a
DOP dioctylphthalate smoke test; DM direct method sterility; LF lactose fermentation (standard methods
of analysis); MLT USP microbial limits test; ORG organic material (oil); PT pyrogen test; V velometer;
VF vial ll or media ll test.
b
B batch or shift; D daily; H history; M monthly.
Table 1.24 Example of Guidelines (Action Levels) for
Environment Monitoring of Sterile Production Elements
Bacteria Mold
Sterile processing location
a
(per cm
2
) (per cm
2
)
Controlled areas (Class 10,000
100,000)
Walls, oors, equipment; swab or 1 1
Rodac
Assembly rooms 10 2
Critical areas (Class 100)
Walls, oors, equipment; swab or 0 0
Rodac
SP SA SP SA
Controlled areasair sampling
b
Component preparation areas 10 100 2 25
Transfer areas 2 25 1 10
Gowning rooms 2 25 1 10
Wash booths 5 50 2 10
Staging areas 50 500 10 100
Compounding rooms 2 25 1 10
Critical areasair sampling
b
Filling rooms 2 10 0 2
Sterility test laboratory 1 1 0 0
Water
c
1/ml 0/ml
Source: From Ref. 83.
a
Controlled areas: sampled following sanitizing and prior to use.
b
Air sampling: settling plate exposed for 30 minutes or slit-to-agar sampling
at a rate of 28.3 liters air/minutes for 30 minutes.
c
Water tests same as for nonsterile manufacturing. Microbial count refers
to at time of use.
The ultimate purpose of environmental control of micro-
bial contamination is to minimize the potential for inadvertent
product contamination. The less the potential for contamina-
tion, the greater the assurance that the product is sterile. The
sterility test then can be used primarily as a conrmation of
the sterility already built into the product.
96 Chapter 1
A 1990 FDA 483 observation was issued to a manufac-
turer who released a lot of product based on the fact that it
passed a sterility test although environmental monitoring
during production of the sterile bulk drug revealed objection-
able conditions. In another case, a manufacturer decided to
release a lot of product based on a successful retest of the ste-
rility test because the contaminant that caused the failure of
the original sterility test was not isolated on any environmen-
tal sample in either the production area or the sterility test
laboratory. However, the FDA countered that environmental
monitoring provides only a snapshot of the environment, and
that it is not surprising that different media and different
sampling times will selectively identify different organisms.
Personnel Training
Most inadvertent contamination found in the sterility testing
of parenteral products originates from the personnel involved
in the testing program. Nearly all personnel-induced acciden-
tal contamination is produced either by the ignorance of an
individual who has not been adequately trained in good asep-
tic technique or by the carelessness of an individual who has
been trained in good aseptic technique. Thus, learning and
applying aseptic technique not only requires physical and in-
tellectual abilities, but also involves the development and per-
sistence of a correct mental attitude. The latter is very difcult
to instill. No one in a free society can be forced to comply to
rigid standards. Supervisors who hire personnel to perform
sterility tests should abide by three general rules:
1. The supervisor must recognize the need to comply with
strict aseptic technique.
2. The supervisor should hire people who are willing to be
trained and to accept and follow aseptic procedures.
3. The supervisor must effectively communicate and exem-
plify the importance of adhering to aseptic technique with-
out breeding ill feelings and subsequent poor attitudes.
Current good manufacturing procedures (CGMPs) (Section
211.25) contain several statements regarding the training of
people engaged in the manufacture, processing, packaging,
and holding of drug products. Personnel will be trained not
only to perform sterility tests, but also to understand CGMPs
and standard operating procedures as they relate to sterility
testing.
Training in correct aseptic technique includes ve gen-
eral areas of education:
1. General rules to followwhen a person is working in a clean
or sterile room
2. Proper gowning technique
3. Proper use of the LAF workbench or other clean environ-
ment
4. Specic operations and manipulations while actually per-
forming the sterility test that are essential in maintaining
asepsis
5. Proper cleanup at the conclusion of the test
No matter how well constructed a sterile or clean room
may be, it cannot compensate for people working in the area
who are untrained with respect to sources of contamination.
DeVecchi (84) has published 29 rules or restrictions concern-
ing when training people to work in sterile environments.
These are listed in Table 1.25.
The importance of gowning may be emphasized best by
reference to a statement by Abdou (85):
A room in which people work cannot be made sterile, re-
gardless of how closely instructions concerning personal
hygiene are followed. Twenty percent of the cutaneous
ora is located so deep within the follicular channels that
it cannot be reached by normal disinfection procedures.
Such a reservoir of organisms ensures that the surface
ora will quickly reestablish itself after the usual treat-
ment of the skin with disinfectants. The epidermal frag-
98 Chapter 1
Table 1.25 General Rules and Procedures for Working
in a Sterile Environment
1. Before entering any sterile environment, personnel should
understand the responsibilitites of their position and know
clean-room techniques and system operations.
2. Personnel must react effectively in emergencies such as: re
outside or inside the sterile room, explosions outside or inside
the sterile room, electrical failure, breaking of containers
holding toxic or nontoxic substances, illness or injury.
3. Everyone who enters the sterile area must be familiar with
gowning technique.
4. Without exception, all personnel working, supervising,
controlling, or maintaining a sterile room should wear the
approved sterile-room garments.
5. No sterile-room garment may be used a second time without
being rewashed and resterilized.
6. Everyone working in sterile areas must know the disinfection
and sterilization procedures.
7. Once inside a sterile room, personnel should avoid returning
to the air lock. If a worker must go to the restroom, complete
resterilization and regarmenting are necessary prior to
reentering the clean room.
8. Plastic bags for disposal of used garments should be provided
in the air locks adjacent to powdered-antibiotic lling or
preparation areas. The garments may be transported in these
bags to the laundry area without risk of cross-contamination
between product and personnel.
9. For reasons of comfort and efciency, establish a minimum
number of people to be allowed in the air locks at any one
time.
10. No personal articles (purses, bags, etc.) are permitted inside
the sterile rooms or air locks.
11. No one who is physically ill, especially with a stomach or
respiratory disorder, may enter sterile rooms or sterile areas.
12. All verbal communication with people outside of the sterile
room should be accomplished through use of the intercom
never through air locks or passthroughs.
13. The sterile-room doors must be kept closed at all times. They
may open only to admit one person or product at a time.
14. Smoking is prohibited inside sterile rooms and neighboring
rooms.
15. The use of cosmetics, wigs, makeup, long nails, rings,
watches, etc., is prohibited in sterile rooms.
16. All materials, containers, or equipment introduced into the
sterile room must be subjected to stringent sterilization
procedures prior to entering the sterile areas.
17. Only long-bered materials may be used for cleaning in
sterile areas. Synthetic materials are suggested. Mops,
brooms, and other customary cleaning equipment should not
be used in sterile areas.
18. Paper in any form (except paper produced expressly for
sterile-room standards, and meeting Class 100 conditions as
delineated in Federal Standard 209B) is not allowed in sterile
rooms.
19. Under no circumstances should food or beverages be
introduced into a sterile room.
20. No pencils or ball-point pens should be used in a sterile room.
Magic markers or felt-tip pens are suggested.
21. When it is necessary that paper forms be used in sterile
areas, the form should be shielded with a clean plastic
covering that has a window exposing the area on which the
operator is writing.
22. Two different products are not to be processed in the same
sterile room at the same time.
23. Antibiotic products in a powder form or liquid products of any
kind should be manufactured in areas designated specically
for that purpose.
24. Disinfection and cleaning of the room must be completed at
scheduled times. All personnel in the sterile room should
know the cleaning and disinfection techniques used.
25. The sterile room must be kept clean at all times. Personnel,
equipment, and materials introduced into a sterile room
should be kept to a minimum.
26. Once production runs are discontinued, any material from the
previous production run should be removed from the sterile
room to avoid cross-contamination.
100 Chapter 1
27. Cleaning and/or disposal of all support material should be
done after each workday.
28. Sterile-room furniture should be of simple design. No chair
covers to chairs with foam parts are allowed in sterile areas.
Tables should be stainless steel and without drawers.
Equipment should be properly covered. No equipment with
belt-driven or high-speed moving parts should be permitted in
a sterile environment unless that equipment has proper
covering.
29. All materials, containers, equipment, etc., authorized for
sterile-room use must be labeled so as to be easily identied
by clean-room personnel.
ments that people shed carry microorganisms, and the
more vigorous the physical activity, the more the shed-
ding. The skin of a healthy adult can shed between two
and six million colony-forming units in one-half hour of
vigorous activity.
Thus, the use of low or particulate gowning materials and ad-
herence to strict gowning procedures will help to ensure that
the human body and clothing will not be a source of contami-
nation. However, Brier et al. (86) reported that employment
of clean room gowning did not affect the contamination rate
of admixtures compounded in a hospital pharmacy. What was
important was that intravenous admixtures were com-
pounded using a LAF workbench.
Proper use of the LAF working environment in the con-
tent of this discussion refers to the movement and manipula-
tions of hands and objects in the hood without interfering or
interrupting the ow of sterile air onto articles that must be
kept sterile. Procedure 8 in Appendix IV should be reiterated
at this point. Opening containers, devices, or other articles in
which a sterile surface or pathway will be exposed should be
completed so that the sterile part faces the HEPA-ltered air.
Moving, tilting, or otherwise manipulating open containers
must be accomplished without ngers and hands either mak-
ing contact with the exposed opening or coming between the
opening and the airstream pathway. Sterility test aids such
as sterile forceps, scissors, lters, and other devices must be
handled with care so as not to touch-contaminate the article.
Whenever the operator suspects that a sterile surface has acci-
dentally been touched, that article should be discarded. Fin-
gers that have been disinfected and subsequently make con-
tact with a nonsterile object should be disinfected again with
a suitable disinfectant solution or foam. The most important
aspect of working in an LAF workbench is mental concentra-
tion on the task at hand, always realizing where the hands
are in relation to the HEPA-ltered airstream and the critical
work sites.
Operator training on the actual sterility testing proce-
dures means the learning of the standard operating procedure
(SOP) written for the sterility test to be executed. This step is
probably the most time-consuming component of the training
process. The operator will work closely with an experienced
supervisor or other trainer for the length of time required for
the operator to learn the SOP and perform the test without
error in procedure and/or technique. The rate of false-positive
sterility test samples will be ascertained for each new opera-
tor; obviously, a certain acceptable rate must be attained for
the operator to be entrusted with future sterility test responsi-
bility. Each sterility testing facility should set up a monitor-
ing program to check periodically the rate of false-positive
samples produced by each operator. Hospital pharmacies that
prepare intravenous admixtures and other sterile products
should also maintain training programs for their sterile prod-
ucts technicians. Organized training programs based on na-
tional standards have been considered by the American Soci-
ety of Hospital Pharmacists (87).
Regardless of how well a person is trained in the proce-
dural aspects of conducting sterility tests, that individual
must also possess the right mental attitude toward the respon-
102 Chapter 1
sibility and implications at hand. Otherwise, a mediocre or
poor attitude will result in carelessness, indifference, and, ul-
timately, errors in technique. A right attitude must be present
in the individual at the beginning and then maintained and
motivated through supervisory encouragement and reward.
ALTERNATIVES TO THE COMPENDIAL STERILITY TEST
The limitations of the USP/NF sterility test have already been
addressed. They include the large sampling error due to the
very small sample size tested for sterility, the problem of inad-
vertent contamination during sterility testing, and the dif-
culty in recovering low-level contamination; these all contrib-
ute to reasons for nding alternative procedures to the USP
test as it is described for a reference test. Another reason for
searching for alternative sterility test procedures is to ll the
need of hospital pharmacies and other laboratory environ-
ments in which sterile solutions are prepared or manipulated
in some manner. The USP referee sterility test is too time con-
suming and costly to be used routinely in hospital practice,
especially with the enormous numbers of intravenous admix-
ture solutions being prepared.
At least two basic methods have been used for sterility
testing in hospital practice. One involves the sampling of an
aliquot volume of solution from an intravenous bottle (45),
while the other method involves the ltration of all of the re-
maining portion of the contents of the bottle through a closed
lter system (88). In the rst method, an aliquot sample is
added to a concentrated broth solution such as double-
strength brain-heart medium, FTM, or other suitable culture
medium, or if it is feasible, the concentrate is added in a vol-
ume equal to the contents of the bottle as it is. The container
is incubated and then inspected for the presence of microbial
growth. The advantage of this method is its simplicity and
cost. Its disadvantages include its potential for accidental con-
tamination and the inability of one culture mediumto promote
the growth of all potential microbial contaminants, especially
in large-volume solutions because of the high dilution factor.
In the second method, a special device, such as Steritest (Milli-
pore; Figure 1.19) or IVEX (Abbott; Fig. 1.20), is designed to
permit the ltration in a closed system of the entire contents
of a bottle through a plastic presterilized unit containing a
0.220.45-m membrane lter. TSB or FTM is then added
aseptically, and the unit is incubated intact. This closed sys-
tem was designed to reduce the rate of false positives and to
Fig. 1.19 Steritest chamber (Millipore Corp., Bedford, Massachu-
setts, from Ref. 78).
104 Chapter 1
Fig. 1.20 IVEX-2 device (courtesy of Abbott Laboratories, North
Chicago, Illinois).
provide a more convenient method of sterility testing large-
volume solutions and admixtures in hospitals (89). Disadvan-
tages of the lter device system are its relatively high cost and
some concern about its sensitivity to low-level contamination.
Posey et al. (90) compared these two methods for de-
tecting microbial contamination in 1-liter plastic bags of par-
enteral nutrition solutions containing 1 and 1000 bacterial or
yeast organisms per milliliter. From each bag, 10-ml aliquots
were withdrawn and injected into blood culture bottles. The
remaining uid was ltered through the Addi-chek system.
The aliquot sampling method consistently detected each of the
organisms tested at levels of 100 organisms per liter and
above. The ltration method consistently detected all levels of
contamination. The authors concluded that the aliquot sam-
pling method was inexpensive and easy to use, but failed to
detect some contaminated solutions. The ltration method de-
tected all levels of contamination, but is more costly in both
time and money, and its reliability needs additional assess-
ment.
The Addi-chek system was compared in a like manner
with the IVEX-2 Filterset in the sterility testing of intrave-
nous admixtures (91). Both lter systems were comparable in
detecting low-level microbial contamination, but the IVEX-2
system can be used to test for contamination when used as an
in-line lter for patient administration of intravenous uids.
Like the aliquot sampling method, the IVEX-2 system is less
expensive than the Addi-chek system. Both IVEX-2 and the
method of combining equal volumes of product sample and
double-strength TSB were found to be more reliable and sensi-
tive than Addi-chek in detecting low-level bacterial contami-
nation in intravenous solutions, especially Dextrose 5%
(92,93). Addi-chek sterility testing consumed more time in
processing, allowing dextrose 5% time to exert an inhibitory
effect on microbial growth. Also, Addi-chek uses a 0.45-m
membrane lter, while IVEX-2 contains a 0.22-m lter.
A modication of the Addi-chek device to be used as an
alternative to the MF method for the sterility testing of antibi-
otics has been described (94). The modied unit is shown in
Fig. 1.18. Two separate spikes are connected to a two-way
valve that prevents the siphoning of the antibiotic into the
rinse-and-media line. The rinse-and-media line is used to
transfer the rinse solution to the canister. The rinse removes
inhibitory residual antibiotic fromthe canister so that the con-
taminating microbes that may have been deposited on the l-
ter may grow. After the rinse procedure, one canister is lled
with TSBand the other with FTM. This systemhas been found
106 Chapter 1
to recover organisms with equal efciency (as compared with
the USP membrane ltration method) and, since it uses a
closed, presterilized, and ready-to-use system, greatly reduces
the chances of operator error and accidental contamination.
The use of an in-line 0.22-m membrane lter set to test
the sterility of intravenous solutions and administration sets
(under actual use conditions) was found to have valuable ap-
plication in intravenous uid administration (95). Following
ltration, brain-heart infusion broth was introduced into the
lter chamber, and the lter sets were incubated. Microbial
contamination was found for all contaminated intravenous so-
lutions and administration sets. No false-positive results were
found.
Sterility testing of small-volume unit dose parenteral
products in the hospital environment was suggested by Rupp
et al. (96), who used a sterile lter unit like those already de-
scribed. For each lot of unit dose syringes, 10% were ltered
before adding FTM to the lter unit. The units were incubated
and inspected for turbidity or color change.
A practical hospital sterility test method for monitoring
intravenous admixtures combines the MF technique and u-
orescent microscopy (97). Solutions are ltered under vacuum
through a 0.6-mmembrane lter. Then, a staining compound
(acridine orange solution) is poured onto the membrane and
allowed to stand for 3 minutes. The stain is removed by vac-
uum, and the membrane is removed and mounted on a micro-
scope slide. The membrane is examined within an hour with
a light microscope tted with an epiuorescent illuminator
system. Any bacteria entrapped on the membrane surface will
react with the uorescent stain, and when illuminated with
incident light, the total number of cells can be counted. The
correlation between uorescent counts and plate colony counts
is excellent, although the counts determined by uorescence
microscopy took only an hour or less, while plate counting re-
quires a 48-hour incubation. Sensitivity of this new method
can be increased to levels as lowas 25 organisms per milliliter,
and the technique can be automated.
Several alternative methods of sterility testing have been
suggested for which neither aliquot sampling nor ltration is
employed. One method suggested the addition of dehydrated
broth powder (thioglycollate medium) to a random selection
of bottles fromeach batch of infusion uids before sterilization
by autoclaving, followed by incubation and daily inspection for
turbidity (98). Another method employed the use of an elec-
tronic particle counter to detect contaminated culture media
within 24 hours after adding contaminated membrane lters
to the media (99). There is also the luciferase assay for adeno-
sine triphosphate (ATP) since detection of ATP indicates the
presence of living cells (100). Each of these proposed sterility
test alternatives offers one or more distinct advantages over
the USP/NF sterility test either in terms of convenience, re-
ducing the incidence of inadvertent contamination, or in
signicantly reducing the time required for detection of con-
taminated products. However, they are not necessarily alter-
natives to be preferred to a referee test in an advisory situa-
tion. Only in certain situations might one of these methods be
a preferable alternative to the USP sterility test, especially in
hospitals. For example, the luciferase ATP assay presents a
very rapid method that can be used in septicemia investiga-
tions in which a large number of intravenous uids must be
tested as quickly as possible.
Several manufacturers expressed the desire to employ an
automated sterility testing system. One commercially avail-
able system is the Bactec system (Johnston Laboratories). Its
principle of operation is based on the detection of radioactive
gas produced by decomposition of labeled substrates by micro-
bial action. Samples of pharmaceutical product are with-
drawn, inoculated into Bactec culture vials containing
14
Csub-
strates, and incubated for 25 days. The Bactec instrument
automatically tests the vials by analyzing the atmosphere in
the vials. If the vial contains microorganisms, they will metab-
olize the
14
C substrates to product
14
CO
2
. A positive result will
be indicated once a threshold level of
14
CO
2
is exceeded. Models
are available to test 60 culture vials per hour. The system of-
108 Chapter 1
fers expediency and convenience not characteristic of ofcial
sterility testing methods.
An article by Lewandoski cites several reasons for the
lack of an approved rapid sterility test (107).
1. Most rapid methods do not allow for subsequent culture
and identication in the event of a positive sterility testing
result. The identication of a positive sterility test result
is key to the investigation (mentioned above).
2. No rapid method is currently sanctioned by any major
pharmacopeia. The USP, EP, and Japanese Pharmacopeia
(JP) all cite the traditional sterility test as the nal release
test for injectable products.
3. Legal issueslack of a compendial sterility test is difcult
to defend in court.
4. False positivessome rapid methods may detect dead
cells.
5. The traditional sterility test has been employed in the
pharmaceutical industry for more than 40 years.
6. To date, no pharmaceutical company has submitted an
NDA with a rapid method as a release test (e.g. no one is
willing to go rst). A representative from CDER is cited
in the article: For a long time, Ive been hoping to see
rapid methods in applications [NDA] where the test is par
of the product release specication. The coming years will
be very interesting as rapid methods develop.
Another alternative to the sterility test is the concept of
parametric release; the sterility test is waived in lieu of a vali-
dated terminal sterilization cycle capable of a high sterility
assurance level. Parametric release can only be employed
when there is assurance that any article processed will meet
the requirement for the sterility test.
Four methods of sterilization that qualify for parametric
release: moist heat (steam), dry heat, ethylene oxide, and ion-
izing radiation. These methods of sterilization are recognized
by U.S. and European regulatory bodies (see USP 1222
and EMEA Note for Guidance on Parametric Release, respec-
tively).
There are several requirements in order for a product to
qualify for parametric release:
1. The manufacturer must obtain authorization from the ap-
propriate regulatory body or bodies prior to initiation.
2. The terminal sterilization method must be designed and
validated to achieve a sterility assurance level of at least
10
6
.
3. The microbial load of the article prior to sterilization
should be quantied and characterized because the micro-
bial load and type(s) of organism(s) present must be con-
sidered for the terminal sterilization cycle.
4. The use of physicochemical indicators is required in every
terminal sterilization load. Indicators should be correlated
to at least a 10
6
log reduction of an appropriate BI (see
Table 1.21 for a list of appropriate BIs for various steril-
ization methods).
5. Verication that the established parameters of the termi-
nal sterilization cycle were performed correctly.
6. Verication that the BIs were reduced to sterility after
sterilization and incubation.
If any of the established parameters are not met or if the bio-
logical indicators are positive after incubation, then the lot
must be rejected. If parametric release is employed, then the
use of the sterility test to justify release of the lot is strictly
forbidden.
REFERENCES
1. C. C. Mascoli, Should end product sterility testing continue?
Med. Device Diag., 3, 89 (1981).
2. F. W. Bowman, The sterility testing of pharmaceuticals, J.
Pharm. Sci., 58, 13011308 (1969).
3. P. M. Borick and J. A. Borick, Sterility of pharmaceuticals,
110 Chapter 1
cosmetics, and medical devices, Quality Control in the Phar-
maceutical Industry, volume 1 (M. S. Cooper, Ed.), Academic
Press, New York, 1972, pp. 138.
4. A. Beloian, Methods of testing for sterility and efcacy of ster-
ilizers, sporicides and sterilizing processes, Disinfection, Ster-
ilization, and Preservation, 2nd Edition (S. S. Block, Ed.), Lea
and Febiger, Philadelphia, 1977, pp. 1148.
5. A. S. Outschoorn, USP standards for sterilization and sterility
testing, Developments in Industrial Microbiology, Volume 18,
Impressions, Inc., Gaithersburg, Maryland, 1977, pp. 387397.
6. W. Olson, Sterility testing, Aseptic Pharmaceutical Manufac-
turing (W. P. Olson and M. J. Groves, Eds.), Interpharm
Press, 1987, pp. 315354.
7. J. E. Akers, F. J. Carleton, W. C. Clements, and J. A. Woods,
Survey on sterility test practices, J. Parenter. Sci. Tech., 41,
197206 (1987).
8. Code of Federal Regulations, Title 21, Section 610.12 for bio-
logics and Section 436.20 for antibiotics, Superintendent of
Documents, U.S. Government Printing Ofce, Washington,
D.C.
9. L. K. Randolph and J. L. Climinera, Statistics, Remingtons
Pharmaceutical Sciences, 16th Edition, Mack Publishing,
Easton, Pennsylvania, 1980, p. 126.
10. Military Standard Sampling Procedure and Tables for Inspec-
tion by Variables for Per Cent Deviation (MIL-STD-414), Su-
perintendent of Documents, U.S. Government Printing Ofce,
Washington, D.C., 1963.
11. S. Lin and L. Lachman, Quality assurance: process and dos-
age form control, The Theory and Practice of Industrial Phar-
macy, 1st Edition (L. Lachman, H. A. Lieberman, and J. L.
Kanig, Eds.), Lea and Febiger, Philadelphia, 1970, p. 734.
12. C. B. Sampson, Quality control of the manufacturing process,
Statistics in the Pharmaceutical Industry (C. R. Buncher and
J. Y. Tsay, Eds.), Marcel Dekker, New York, 1981, p. 335.
13. J. Brewer, A clear liquid medium for the aerobic cultivation
of anaerobics, J. Bacteriol., 39, 10 (1940).
14. J. E. Doyle, W. H. Mehrhof, and R. R. Ernst, Limitations of
thioglycollate broth as a sterility test medium for materials
exposed to gaseous ethylene oxide, Appl. Micro., 16, 1742
1744 (1968).
15. F. W. Bowman, M. White, and M. P. Calhoun, Collaborative
study of aerobic media for sterility testing by membrane l-
tration, J. Pharm. Sci., 60, 10871088 (1971).
16. M. A. Abdou, Comparative study of seven media for sterility
testing, J. Pharm. Sci., 63, 2326 (1974).
17. F. W. Bowman, Application of membrane ltration to antibi-
otic quality control sterility testing, J. Pharm. Sci., 55, 818
821 (1966).
18. A. G. Mathews, Optimal incubation conditions for sterility
tests, Dev. Biol. Stand., 23, 94102 (1974).
19. M. Pittman and J. C. Feeley, Proc. 7th International Congress
Microbiol. Standardization, 7th Edition, E and S Livingstone,
Edinburgh and London, 1962.
20. R. L. DeChant, D. Furtado, F. M. Smith, H. N. Godwin, and
D. E. Domann, Determining a time frame for sterility testing
of intravenous admixtures, Am. J. Hosp. Pharm., 39, 1305
1308 (1982).
21. M. P. Calhoun, M. White, and F. W. Bowman, Sterility testing
of insulin by membrane ltration: a collaborative study, J.
Pharm. Sci., 59, 10221024 (1970).
22. G. W. Riedel and E. C. D. Todd, Sterility testing of large-vol-
ume aqueous pharmaceutical products (intravenous solution)
by the membrane ltration technique, Can. J. Pharm. Sci., 8,
2325 (1973).
23. G. G. Christiansen and T. A. Koski, J. Biol. Stand., 11, 83
89 (1983).
24. H. L. Avallone, Sterility retesting, J. Parenter. Sci. Tech., 40,
5657 (1986).
25. J. Y. Lee, Investigating sterility test failures, Pharm. Tech.,
14, 3841 (1990).
26. D. A. Quagliaro, G. T. Spite, D. E. Guilfoyle, and W. E. Mes-
trande, Development of culture vessel for sterility testing of
other medical devices, J. Assoc. Off. Anal. Chem., 62, 695
699 (1979).
112 Chapter 1
27. W. T. Sokolski and C. G. Chidester, Improved viable counting
method for petrolatum-based ointments, J. Pharm. Sci., 53,
103107 (1964).
28. K. Tsuji, E. M. Stapert, J. H. Robertson, and P. M. Waiyaki,
Sterility test method for petrolatum-based ophthalmic oint-
ments, Appl. Microbiol., 20, 798801 (1970).
29. K. Tsuji and J. H. Robertson, Microbiol toxicity of isopropyl
myristate used for sterility testing of petrolatum-based oph-
thalmic ointments, Appl. Microbiol., 25, 139145 (1973).
30. A. M. Placenci, G. S. Oxborrow, and J. W. Davidson, Sterility
testing of fat emulsions using membrane ltration and di-
methylsulfoxide, J. Pharm. Sci., 71, 704 (1982).
31. J. H. Brewer and R. F. Schmitt, Special problems in the steril-
ity testing of disposable medical devices, Bull. Parenter. Drug.
Assoc., 21, 136141 (1967).
32. L. W. Gee, J. M. G. H. Harvey, W. P. Olson, and M. L. Lee,
Sterility test systems for product recovery, J. Pharm. Sci., 74,
2932 (1985).
33. R. R. Ernst, K. L. West, and J. E. Doyle, Problem areas in
sterility testing, Bull. Parenter. Drug Assoc., 23, 2939 (1969).
34. L. F. Knudsen, Sample size of parenteral solutions for sterility
testing, J. Am. Pharm. Assoc., Sci. Ed., 38, 332337 (1949).
35. J. H. Brewer, Antiseptics, Disinfectants, Fungicides, and Ster-
ilization, 2nd Edition (G. L. Reddish, Ed.), Lea and Febiger,
Philadelphia, 1957, pp. 160161.
36. Proc. Round Table Conf. Sterility Testing, London, 1963,
p. B31.
37. P. Armitage, A note on the safety testing of vaccines, J. Hyg.
(Lond.), 69, 9597 (1971).
38. J. L. Friedl, Report on sporicidal tests, J. Assoc. Off. Agric.
Chem., 38, 280287 (1955).
39. D. A. A. Mossel and H. Beerens, Studies on the inhibitory
properties of sodium thioglycollate on the germination of wet
spores of Clostridia, J. Hyg., 66, 269272 (1968).
40. C. W. Bruch, Levels of sterility: probabilities of survivors ver-
sus biological indicators, Bull. Parenter. Drug Assoc., 28, 105
121 (1974).
41. I. Scheibel and M. W. Bentzon, In Proc. Third Int. Meeting
of Biological Standardization, Opatija, Biostandards, Geneva,
1957.
42. A. N. Parisi and P. M. Borick, Proc. 7th Annu. Mtg. Assoc.,
Contam. Contr., 7, 2426 (1968).
43. C. J. Luna, Personnel: the key factor in clean roomoperations,
Pharmaceutical Dosage Forms: Parenteral Medications, Vol-
ume 2 (K. E. Avis, H. A. Lieberman, and L. Lachman, Eds.),
Marcel Dekker, New York 1992.
44. National Coordinating Committee on Large-Volume Paren-
terals, Recommended procedures for in-use testing of large-
volume parenterals suspected of contamination or of produc-
ing a reaction in a patient, Am. J. Hosp. Pharm., 35, 687682
(1978).
45. J. J. Bernick, D. G. Brown, and J. E. Bell, Adventitious con-
tamination of intravenous admixtures during sterility test-
ing, Am. J. Hosp. Pharm., 36, 14931496 (1979).
46. A. J. Hanson, R. M. Nighswander, and J. H. Verhulst, Moni-
toring of intravenous solutions, Hosp. Formul., 8, 1721
(1973).
47. J. Whitbourne and K. West, Sterility testing: how appropriate
for the hospital? Med. Instrum., 10, 291292 (1976).
48. W. A. Zellmer, Quality control in admixture services editorial,
Am. J. Hosp. Pharm., 35, 527528 (1978).
49. Recommendations to pharmacists for solving problems with
largevolume parenterals1979, Am. J. Hosp. Pharm, 37,
663667 (1980).
50. M. H. Stolar, Assuring the quality of intravenous admixture
programs, Am. J. Hosp. Pharm., 36, 605 (1979).
51. National Coordinating Committee on Large-Volume Paren-
terals, Recommended guidelines for quality assurance in hos-
pital centralized intravenous admixture services, Am. J.
Hosp. Pharm., 35, 678 (1978) and 37, 645 (1980).
52. A. L. Hanson and R. M. Shelly, Monitoring contamination of
114 Chapter 1
inuse intravenous solutions using total sample techniques,
Am. J. Hosp. Pharm., 31, 733735 (1974).
53. J. A. Buth, R. W. Coberly, and F. M. Eckel, A practical method
of sterility monitoring of I.V. admixtures and a method of im-
plementing a routine sterility monitoring program, Drug In-
tell. Clin. Pharm., 7, 276279 (1973).
54. R. B. Kundsin, C. W. Walter, and J. A. Scott, Inuse testing
of sterility of intravenous solutions in plastic containers, Sur-
gery, 73, 778781 (1973).
55. R. Ravin, J. Bahr, F. Luscomb, et al., Program for bacterial
surveillance of intravenous admixtures, Am. J. Hosp. Pharm.,
31, 340347 (1974).
56. L. H. Sanders, S. A. Mabadcje, K. E. Avis, et al., Evaluation of
compounding accuracy and aseptic techniques for intravenous
admixtures, Am. J. Hosp. Pharm., 35, 531536 (1978).
57. N. C. d Arbeloff, S. Rusmin, and D. Stack, Robotic applica-
tions in sterility testing, Sterile Pharmaceutical Manufactur-
ing (J. V. Groves, W. P. Olson, and M. H. Ansfeld, Eds.), In-
terpharm Press, 1991, pp. 199220.
58. B. J. Zlotnick and M. L. Franklin, A robotic system for the
sterility testing of injectables, Pharm. Tech., 11, 5964 (1987).
59. S. M. Davenport, Design and use of novel peracetic acid steril-
izer for absolute barrier sterility testing chambers, J. Pare-
nter. Sci. Tech., 43, 158166 (1989).
60. M. S. Cooper, Microbiol. Update, 8(7) (1990).
61. P. L. Simmons, The secret of successful sterilizer validations
(part 13), Pharm. Eng., 1 (Nov. 1980July 1981).
62. A. R. Reti and T. J. Leahy, Validation of bacterially retentive
lters by bacterial passage testing, J. Parenter. Drug Assoc.,
33, 257272 (1979).
63. M. J. Akers, I. A. Attia, and K. E. Avis, Understanding and
utilizing F
0
values, Pharm. Tech., 2, 3135 (1978).
64. J. B. Selkon, P. R. Sisson, and H. R. Ingham, The use of spore
strips for monitoring the sterilization of bottled uids, J. Hyg.
(Lond.), 83, 121125 (1979).
65. D. H. Dadd and G. M. Daley, Resistance of microorganisms
to inactivation by gaseous ethylene oxide, J. Appl. Bacteriol.,
49, 89101 (1980).
66. P. M. Borick and J. A. Borick, Sterility of pharmaceuticals,
cosmetics, and medical devices, Quality Control in the Phar-
maceutical Industry, Volume 1 (M. S. Cooper, Ed.), Academic
Press, New York, 1972, p. 17.
67. T. Myers and S. Chrai, Parenteral fundamentals: basic princi-
ples and applications of bioindicators, J. Parenter. Drug
Assoc., 34, 234243 (1980).
68. R. A. Caputo and C. C. Mascoli, The design and use of biologi-
cal indicators for sterilization cycle validation, Med. Device
Diagn., 2, 23 (1980).
69. I. J. Pug and T. E. Odlaug, Syllabus for an Introductory
Course in the Microbiology and Engineering of Sterilization
Processes, Environmental Sterilization Services, St. Paul,
Minnesota, Aug. 1978.
70. R. A. Caputo, T. E. Odlaug, R. L. Wilkinson, and C. C. Mascoli,
Biological validation of a steam-sterilized product by the frac-
tional exposure method, J. Parenter. Drug Assoc., 33, 214
221 (1979).
71. Federal Standard 209E, Clean Room and Work Station Re-
quirements, Controlled Environment, General Services Ad-
ministration, Washington, D.C., Sept. 11, 1992.
72. W. Whiteld, Contam. Contr., 3, 16 (1964).
73. G. B. Phillips and W. S. Miller, Sterilization, Remingtons
Pharmaceutical Sciences, 16th Edition, Mack Publishing,
Easton, Pennsylvania, 1980, pp. 13991400.
74. Personal communication, K. E. Avis, College of Pharmacy,
University of Tennessee Center for the Health Sciences, Mem-
phis, Tennessee.
75. R. I. Gross, Laminar ow equipment performance and testing
requirements, Bull. Parenter. Drug Assoc., 30, 143151
(1976).
76. R. I. Gross, Testing for laminar ow equipment, J. Parenter.
Drug. Assoc., 32, 174181 (1978).
77. ASTM D2986-71, Standard method for evaluation of air assay
media by the monodisperse DOP smoke test.
116 Chapter 1
78. G. B. Phillips and R. S. Runkle, Biomedical Applications of
Laminar Airow, CRC Press, Cleveland, Ohio, 1973.
79. K. E. Avis, Parenteral preparations, Remingtons Pharmaceu-
tical Sciences, 16th Edition, Mack Publishing, Easton, Penn-
sylvania, 1980, pp. 14701475.
80. R. R. Austin, Encyclopedia of Clean Rooms/White Rooms/
Sterile Rooms, Contamination Control Seminars, Livonia,
Michigan, 1991.
81. R. P. Delmore and W. N. Thompson, Acomparison of airsam-
pler efciencies, Med. Device Diagn., 3, 45 (1981).
82. G. Kraidman, The microbiology of airborne contamination
and air sampling, Drug Cosmet. Ind., 116, 4045 (1975).
83. L. A. Dell, Aspects of microbiological monitoring for nonster-
ile and sterile manufacturing environments, Pharm. Tech., 3,
4751 (1979).
84. F. A. DeVecchi, Training personnel to work in sterile environ-
ments, Pharm. Tech., 2, 4144 (1978).
85. M. A.-F. Abdou, Determination of airborne microorganisms in
a pharmaceutical plant using standard elective and selective
culture media, Pharm. Tech., 4, 93100 (1980).
86. K. L. Brier, C. J. Latiolais, P. J. Schneider, et al., Effect of
laminar air ow and cleanroom dress on contamination
rates of intravenous admixtures, Am. J. Hosp. Pharm., 38,
11441147 (1981).
87. W. A. Gouvia and E. R. Anderson, Protecting our investment
in sterile products, Am. J. Hosp. Pharm., 37, 1311 (1980).
88. B. L. Green and W. Litsky, Evaluation of a closed system for
sterility testing of parenterals, Pharm. Tech., 3, 7277 (1979).
89. C. G. Mayhall, P. G. Pierpaoli, G. O. Hall, and R. B. Thomas,
Evaluation of a device for monitoring sterility of injectable
uids, Am. J. Hosp. Pharm., 38, 11481150 (1981).
90. L. M. Posey, R. E. Nutt, and P. D. Thomson, Comparison of
two methods for detecting microbial contamination in intrave-
nous uids, Am. J. Hosp. Pharm., 38, 659662 (1981).
91. F. Condella, K. Eichelberger, L. C. Foote, and R. E. Grifn,
Evaluation of two sterility testing methods for intravenous
admixtures, Hosp. Pharm., 15, 305310 (1980).
92. K. H. Hoffman, F. M. Smith, H. N. Godwin, L. C. Hogan, and
D. Furtado, Evaluation of three methods for detecting bacte-
rial contamination in intravenous solutions, Am. J. Hosp.
Pharm., 39, 12991302 (1982).
93. C. M. Miller, D. Furtado, F. M. Smith, H. N. Godwin, L. C.
Hogan, and D. E. Letendre, Evaluation of three methods for
detecting lowlevel bacterial contamination in intravenous
solutions, Am. J. Hosp. Pharm., 39, 13021305 (1982).
94. M. Pickett and L. Litsky, An effective alternative for testing
antibiotic sterility, Pharm. Tech., 5, 6368 (1981).
95. J. C. Lim, Technique of microbiological testing of inuse in-
travenous solutions and administration sets, Am. J. Hosp.
Pharm., 36, 12021204 (1979).
96. C. A. Rupp, C. A. Kikugawa, S. E. Kotabe, et al., Quality con-
trol of smallvolume sterile products, Am. J. Hosp. Pharm.,
34, 4749 (1977).
97. S. P. Deniser and K. H. Ward, A rapid method for the detec-
tion of 59 bacterial contaminants in intravenous uids using
membrane ltration and epiuorescence microscopy, J. Pare-
nter. Sci. Tech., 37, 156158 (1983).
98. J. A. Rycroft and D. Moon, An inproduction method for test-
ing the sterility of infusion uids, J. Hyg. Camb., 74, 1725
(1975).
99. T. R. Rameshbabu and A. S. Arambulo, Early detection of mi-
crobial contamination in sterile solutions using an electronic
counter, Bull. Parenter. Drug Assoc., 30, 8087 (1976).
100. C. A. Bopp and I. K. Wachsmuth, Luciferase assay to detect
bacterial contamination of intravenous uids, Am. J. Hosp.
Pharm., 38, 17471750 (1981).
101. M. S. Cooper, Microbiol. Update, 8(1) (January 1990).
102. J. Lysfjord and M. Porter, Barrier isolation history and
trends, Int. Soc. Pharm. Eng., 8086 (Sept.Oct. 1998).
103. FDA proposes new sterility requirements for inhalation solu-
tions for nebulization, Int. Pharm. Reg. Monit., 4, (October
1997).
118 Chapter 1
104. M. Kokubo, T. Inoue, and J. Akers, Resistance of common en-
vironmental spores of the genus Bacillus to vapor hydrogen
peroxide, PDA J. Pharm. Sci. Technol. 228231 (September
October 1998.
105. H. Bathgate, D. Lazzari, H. Cameron, and D. McKay, The in-
cubation period in sterility testing, PDA J. Pharm. Sci. Tech-
nol. 254257 (SeptemberOctober 1993).
106. J. Wilson and M. Varney, Sterility test incubation issues, PDA
J. Pharm. Sci. Technol., 157159 (JulyAugust 1995).
107. N. Lewandoski, Is there time for rapid micro? Pharm. Form.
Qual. 1722 (April/May 2001).
2
Pyrogen Testing
INTRODUCTION
When injected into humans in sufcient amounts, pyrogens
will cause a variety of adverse physiological responses (Table
2.1). The most common or recognizable response is an increase
in body temperature, from which the name pyrogen is de-
rived (Greek pyro re, gen beginning). Pyrogenic re-
sponses rarely are fatal unless the patient is very sick and the
dose is very large. Nevertheless, pyrogens are considered toxic
substances and should never be injected knowingly. Pyrogen
contamination of large-volume parenteral (LVP) solutions is
especially serious because of the large amounts of uid admin-
istered to people whose illnesses must be severe enough to
warrant the use of such large volumes.
Pyrogens come from microorganisms. All microbial forms
produce pyrogen; however, the most potent pyrogen originates
from gram-negative bacteria. The entity primarily involved in
pyrogenic reactions in mammals is the lipopolysaccharide
(LPS) from the outer cell membranes of gram-negative bacte-
119
120 Chapter 2
Table 2.1 Adverse Physiological Effects of Pyrogens in Humans
Primary
Increase in body temperature
Chilly sensation
Cutaneous vasoconstriction
Pupillary dilation
Piloerection
Decrease in respiration
Rise in arterial blood pressure
Nausea and malaise
Severe diarrhea
Pain in the back and legs
Headache
Secondary
Cutaneous vasodilation
Hyperglycemia
Sweating
Fall in arterial blood pressure
Involuntary urination and defecation
Decreased gastric secretion and motility
Penile erection
Leucocytopenia, leucocytosis
Hemorrhage and necrosis in tumors
Altered resistance to bacterial infections
Depletion of liver glycogen
Rise in blood ascorbic acid
Rise in blood nonprotein nitrogen and uric acid
Decrease in plasma amino acids
ria (1). Another name for LPS is endotoxin. Although not en-
tirely correct, the names pyrogen, LPS, and endotoxin are rou-
tinely used interchangeably. Figure 2.1 is a schematic
representation of the three cell wall layers of a gram-negative
microorganism (1). The outer membrane shown in the gure
is not found in gram-positive bacteria. This structure contains
the LPS moiety that interacts with the coagulable protein of
Pyrogen Testing 121
Fig. 2.1 Schematic representation of the three cell wall layers of
a gram-negative bacterium. (From Ref. 1).
the amebocytes of the horseshoe crab, a phenomenon from
which evolved the Limulus Amebocyte Lysate (LAL) test.
LPS, extracted and recovered as a colloidal suspension,
may be split by mild acid hydrolysis into lipid A and degraded
polysaccharides (2). Lipid A is composed of B-1, 6-glucosamine
disaccharide units with a -hydroxymyristic acid replacing one
of the amino hydrogens and fatty acids replacing hydrogen in
some of the EOH groups (see Fig. 2.2). Each two glucosamine
units are separated by two phosphate moieties, forming a lin-
ear polymer (1). Lipid A alone lacks biologic activity, yet LPS
is toxic, probably because polysaccharide increases the aque-
ous solubility of lipid A. Kennedi et al. (3) showed that when
lipid A is separated from the polysaccharide component of en-
dotoxin, it loses more than 99.9% of its pyrogenic activity in
rabbits.
Freedom from pyrogenic contamination characterizes
parenteral products in the same manner as sterility and free-
dom from particulate matter. Preventing the presence of pyro-
gens is much preferred over removing pyrogens in parenteral
products. Preventing pyrogenic contamination primarily in-
volves the use of ingredients, solvents, packaging materials,
122 Chapter 2
Fig. 2.2 Structure of unit of lipid A from Salmonella lipopolysac-
charide. KDO: 3-deoxy-D-mannooctulosonic acid; HM: a -hydroxy-
myristic acid; FA: other long-chain fatty acids. (Adapted from Ref.
2 and Reitschel et al., Eur. J. Biochem., 28, 166, 1972).
and processing equipment that have been depyrogenated ini-
tially, then employing correct and proper procedures during
the entire manufacturing process to minimize the possibility
of pyrogen development.
HISTORY
The pyrogenic response has been known since 1865, when it
was reported that an injection of distilled water produced hy-
perthermia in dogs (4). Later, in 1876, the presence of a fever-
producing substance, called a pyrogen for the rst time, was
found in extracts of putrefying meat (5). Identication of the
pyrogenic component from bacteria was attempted by Roussy
in 1889 (6) and Centanni in 1894 (7), who determined that the
pyrogen was nonproteinaceous. Hort and Penfold (8) in 1911
made signicant contributions in relating the production of
fever and the administration of intravenous infusions. They
also were the rst to use rabbits as an animal model to study
Pyrogen Testing 123
the pyrogenic response. They showed that the incidence of
chills and fever following intravenous injection could be re-
duced markedly if freshly prepared distilled water was used
as the injection solvent. Investigators (9,10) related fever pro-
duction in rabbits with the injection of bacterial culture ex-
tracts and showed that sterile solutions free from endotoxins
did not cause the febrile response. Pyrogenicity seemed to be
related to the gram stain reaction; gram-negative organisms
produced a pyrogenic response, while gram-positive organ-
isms did not. In addition, bacterial pyrogens were not de-
stroyed by autoclaving or removed by ltration.
It is interesting to note that while the medical signi-
cance of a pyrogen was recognized during these years, it was
not until 1923 that Seibert (11,12) recommended that all phar-
maceuticals be tested for pyrogens. Seiberts carefully con-
trolled experiments conrmed Hort and Penfolds (8) results
using the rabbit as the animal model for detecting the pres-
ence of pyrogens in injectables. Seibert also demonstrated con-
clusively that pyrogens originate fromwater-borne organisms,
and are heat resistant, lterable, and can be eliminated from
water by distillation. Rademacher (13) substantiated Sieberts
results and presented instructions for the preparation of pyro-
gen-free parenteral solutions. CoTui and Schrift (14) reported
that the pyrogen-producing characteristics of microorganisms
depend on the type of organism, and that bacterial pyrogens
are related to lipopolysaccharides.
The pyrogen test became an ofcial quality control test
for parenterals in 1942 in the U.S. Pharmacopeia (USP), 12th
edition. Later, in 1945, the Code of Federal Regulations
(CFR)* required antibiotics to be tested for pyrogens. Despite
the advances in parenteral science and technology over the
past 50 years, the rabbit pyrogen test methodology ofcially
recognized in compendial standards has remained essentially
* CFR Title 21, Section 610.13 for biologicals and Sections 436.31 and 436.32 for anti-
biotics.
124 Chapter 2
unchanged. The LAL test for endotoxin, discussed later, be-
came an ofcial USP test in 1985. Today, the LAL test, more
commonly called the Bacterial Endotoxin Test, has preempted
the rabbit test as the USP method of choice for detection of
endotoxin in parenteral products. The traditional rabbit pyro-
gen test has been almost completely replaced by the more sen-
sitive and accurate LAL assay for testing raw materials, in-
process pyrogen control for pharmaceuticals and medical de-
vices, and end-product evaluation of devices, is small and
large-volume parenteral products (15). The LAL test also is
widely used in the validation of depyrogenation of dry-heat
sterilization processes.
SPECIFIC REQUIREMENTS OF THE USP RABBIT
PYROGEN TEST
Since its inception in the USP in 1942, the rabbit pyrogen test
has remained essentially unchanged. Thus, this section fol-
lows closely both the specications written in the 22nd edition
of the USP (16) and the excellent review article written in
1973 by Peroneus (17). The majority of the parenteral industry
relies on the LAL test for assurance of an endotoxin free prod-
uct. The discussion of rabbit testing below is included for his-
torical data. For biologics, some countries (e.g., Canada) still
require the use of the rabbit pyrogen test.
General Description of the USP Pyrogen Test
The following paragraph is quoted directly from the USP
85 under the section on pyrogen testing:
The pyrogen test is designed to limit to an acceptable
level the risks of febrile reaction in the patient to the ad-
ministration, by injection, of the product concerned. The
test involves measuring the rise in temperature of rabbits
following the intravenous injection of a test solution and
is designed for products that can be tolerated by the test
rabbit in a dose not to exceed 10 ml per kg injected intra-
Pyrogen Testing 125
venously within a period of not more than 10 minutes.
For products that require preliminary preparation or are
subject to special conditions of administration, follow the
additional directions given in the individual monograph
or, in the case of antibiotics or biologics, the additional
directions given in the federal regulations.
Apparatus and Diluents
All apparatusesglassware, containers, syringes, needles,
and soonand all diluents used in performing the pyrogen
test must themselves be free from pyrogenic contamination.
Heat-durable items such as glass and stainless steel can be
depyrogenated by exposure to dry heat cycles at temperatures
greater than 250C for at least 60 minutes. Diluents and solu-
tions for washing and rinsing of devices are to be pyrogen free.
Commercially available sterile and pyrogen-free solution
products usually are employed.
To ensure the lack of pyrogenicity with the various mate-
rials used in conducting the pyrogen test, negative controls
should be performed with each test. Negative controls utilize
the diluent rather than the product sample as the injection,
with the diluent being exposed to the same procedure and ma-
terials as the product sample. The use of negative controls
with each pyrogen test is not standard practice because of
prior knowledge and assurance that materials used in the test
are nonpyrogenic.
Temperature Recording
USP 85 states the following:
Use an accurate temperature-sensing device such as a
clinical thermometer, or thermistor probes or similar
probes that have been calibrated to assure an accuracy
of 0.1C and have been tested to determine that a maxi-
mum reading is reached in less than 5 minutes. Insert
the temperature-sensing probe into the rectum of the test
rabbit to a depth of not less than 7.5 cm, and, after a pe-
126 Chapter 2
riod of time not less than that previously determined as
sufcient, record the rabbits body temperature.
Thermocouples connected to electronic recording devices
are almost exclusively used today for measuring temperature
rectally in rabbits. A thermocouple contains two dissimilar
electrical conductor wires joined at one end to form a measur-
ing junction that produces a thermal electromotive force
(EMF). There exist several thermocouple types, each having
a dened EMF-temperature relationship. For example, at a
temperature of 100F, a type T (copper-constantan) thermo-
couple will generate an EMF of 1.518 mV. Common thermo-
couple types are listed in Table 2.2. Typical thermocouples are
composed of three parts (Fig. 2.3). The two dissimilar wires
are supported by an electrical insulator, either hard-red ce-
ramic or nonceramic materials such as Teon, polyvinyl chlo-
ride, berglass, brous silica, or asbestos. The outer sheath
can be composed of a variety of materials, most commonly
stainless steel, Teon, and various elemental metals (plati-
num, copper, and aluminum).
Thermocouples must be accurately calibrated against
National Institute of Standards and Technology (NIST) trace-
able standard constant temperature baths. Accuracy of ther-
mocouple temperature measurement can never exceed the
accuracy of the thermocouple reference. Reference instrumen-
tation should include both an ice point reference bath and an
Table 2.2 Commonly Used Thermocouple Types
Type B Platinum30% rhodium () versus platinum6% rho-
dium ()
Type E Nickel10% chromium () versus constantan ()
Type J Iron () versus constantan ()
Type K Nickel10% chromium () versus nickel5% ()
Type R Platinum13% rhodium () versus platinum ()
Type S Platinum10% rhodium () versus platinum ()
Type T Copper () versus constantan ()
Pyrogen Testing 127
Fig. 2.3 Composition of a typical thermocouple used in rabbit pyro-
gen testing. (Courtesy of the American Society for Testing and Mate-
rials, Philadelphia, Pennsylvania).
elevated temperature reference bath. These calibration baths
initially should be calibrated against an electronic monitor in-
corporating an NBS-traceable standard resistor with an accu-
rate and constant source of current. Once the baths are cali-
brated, the thermocouples can be placed in the wells of the
baths and temperature accuracy determined. The accuracy of
the thermocouples must be 0.1Cof the calibration bath tem-
perature or they should not be used in pyrogen testing.
Rabbit body temperature data are recorded electronically
by instruments such as those seen in Fig. 2.4. Electronic tem-
perature recorders usually can monitor over 100 rabbits si-
multaneously. Any variation in room temperature must be
compensated by built-in calibration capability of the recorder.
Proper maintenance and repair of recording devices must be
accomplished.
Computerized equipment is now available for automatic
temperature recording during pyrogen testing. A description
of computerized temperature recording in pyrogen testing was
published by Joubert (18).
Test Animals
Rabbits are used as pyrogen test models because they physio-
logically respond similarly to pyrogens as do human beings.
128 Chapter 2
Fig. 2.4 Electronic thermal recording instruments used to monitor
rabbit body temperatures during the pyrogen test.
Griesman and Hornick (19) showed that rabbits and humans
respond identically on a nanogram per kilogram basis to pyro-
genic quantities of endotoxin.
Quoting from the USP:
Use healthy, mature rabbits. House the rabbits individu-
ally in an area of uniform temperature between 20C and
23C and free from disturbances likely to excite them.
The temperature varies not more than 3C from the se-
lected temperature. Before using a rabbit for the rst
time in a pyrogen test, condition it not more than seven
days before use by a sham test that includes all of the
steps as directed under Procedure except injection. Do not
use a rabbit for pyrogen testing more frequently than
once every 48 hours, nor prior to 2 weeks following a max-
imum rise of its temperature of 0.6C or more while being
Pyrogen Testing 129
subjected to the pyrogen test or following its having been
given a test specimen that was adjudged pyrogenic.
Several strains of rabbits are acceptable as test animals
for the pyrogen test. Key factors in selecting rabbits are the
animal breeder, rabbit resistance to disease, sufcient size for
ease of handling, large ears, and rate of weight gain. The al-
bino rabbit is the most widely used rabbit, particularly strains
from New Zealand and Belgium.
It is essential that the rabbit colony be treated with ut-
most care. The environment in which the rabbits are housed
must be strictly controlled with respect to temperature, hu-
midity, lighting, and potential contamination of air, surfaces,
and feed. Any new shipment of rabbits should be quarantined
and monitored for 1 to 2 weeks following receipt of the ship-
ment for presence of illness and/or disease.
Rabbits must be trained to adjust and adapt to their new
environment in the pyrogen testing laboratory. Methods ap-
plied have been reviewed by Personeus (17). Rabbits must be-
come accustomed to being restrained in their cages and being
handled during both the rectal insertion of the thermocouple
and the injection of the test product.
The normal basal body temperature of rabbits ranges be-
tween 38.9C and 39.8C (102.0103.6F). Rabbit baseline
temperature is established by measuring rectal temperature
during the conductance of several sham tests (following the
entire pyrogen test procedure using pyrogen-free sodium chlo-
ride solution as the injection sample). Such tests should be,
but rarely are, conducted over a period of several weeks. Tem-
perature variances will occur in untrained rabbits, but on
training, temperature variation will diminish to an acceptable
range of 0.2C. The normal temperature range of a rabbit
may shift with time, requiring the reestablishment of the true
normal body temperature.
Rabbits may become tolerant to pyrogenic activity after
repeated injections of endotoxin (2022). It is for this reason
130 Chapter 2
that a rabbit showing a rise of its body temperature of 0.6C
or more during a pyrogen test cannot be used again as a pyro-
gen test animal for at least 2 weeks.
Test Procedures
The USP procedure recommended for performing the pyrogen
test is as follows:
Perform the test in a separate area designated solely for
pyrogen testing and under environmental conditions sim-
ilar to those under which the animals are housed and free
from disturbances likely to excite them. Withhold all food
fromthe rabbits used during the period of the test. Access
to water is allowed at all times, but may be restricted dur-
ing the test. If rectal temperature-measuring probes re-
main inserted throughout the testing period, restrain the
rabbits with light-tting stocks that allow the rabbits to
assume a natural resting posture. Not more than 30 min-
utes prior to the injection of the test dose, determine the
control temperature of each rabbit. This is the base for
the determination of any temperature increase resulting
from the injection of a test solution. In any one group of
test rabbits, use only those rabbits whose control temper-
atures do not vary by more than 1 degree fromeach other,
and do not use any rabbit having a temperature ex-
ceeding 39.8C.
Unless otherwise specied in the individual mono-
graph, inject into an ear vein of each of three rabbits 10
ml of the test solution per kg of body weight, completing
each injection within 10 minutes after the start of admin-
istration. The test solution is either the product, consti-
tuted if necessary as directed in the labeling, or the mate-
rial under test treated as directed in the individual
monograph and injected in the dose specied therein. For
pyrogen testing of devices or injection assemblies, use
washings or rinsings of the surfaces that come in contact
with the parenterally-administered material or with the
injection site or internal tissues of the patient. For exam-
Pyrogen Testing 131
ple, 40 ml of sterile, pyrogen-free saline, TS at a ow rate
of approximately 10 ml per minute is passed through the
tubing of each of 10 infusion assemblies. Assure that all
test solutions are protected from contamination. Perform
the injection after warming the test solution to a temper-
ature of 37 2C. Record the temperature at 1 and 3
hours and 30 minute intervals in between subsequent to
the injection.
Rabbits belong in a facility that is temperature con-
trolled, for example, at 70C5F. Housing should be individ-
ual cages designed to maintain cleanliness (see Fig. 2.5). Cage
design should conform to standards established by the Ameri-
can Association of Accreditation of Laboratory Animal Care
(AAALAC).
The facility has two basic rooms. One room houses the
rabbits between tests, while the other room is used only for
Fig. 2.5 Housing of pyrogen test rabbits in clean, individual cages.
132 Chapter 2
Fig. 2.6 Rabbits situated in individual restraining boxes.
actual pyrogen testing. Rabbits in restraining boxes (see Fig.
2.6) are transported on carts or wagons from the holding room
into the testing room. The two rooms should have a door be-
tween them that is closed during the pyrogen testing period.
Environmental conditions in the two rooms should be iden-
tical.
Noise represents a major problem in maintaining and us-
ing rabbits for pyrogen testing. The room in which the tests
are conducted should be as free from noise and activity as pos-
sible. Anything that causes excitement in the rabbit poten-
tially can produce a 0.21.0C rise in body temperature that
may not return to normal for 6090 minutes.
During the pyrogen test, which could last 4 to 6 hours, the
rabbits should be restrained with a minimum of discomfort.
Restraint should be conned to the neck and head of the rabbit
to facilitate the test dose injection into the ear vein and to
permit the rabbit comfortable movement of its legs and back.
Pyrogen Testing 133
Fig. 2.7 Rear view of rabbits in restraining boxes.
Examples of modern restraining boxes are shown in Figs. 2.6
and 2.7.
Rabbits that have been adequately trained, are healthy,
and exhibit stable body temperatures are selected for the pyro-
gen test. The animals are weighed and placed in their re-
straining boxes. Thermocouples (see Fig. 2.7) are inserted in
the rectum to a depth of not less than 7.5 cm. Following a 30
45-minute acclimation period, the control temperature read-
ing of the rabbit is recorded. Within 30 minutes of the record-
ing of the control temperature, the test dose should be admin-
istered.
Dose administration is accomplished using a sterile sy-
ringe and 2023-gauge needle. The size of syringe will depend
on the dose volume. The USP requires a dose of 10 ml per
kg body weight unless otherwise specied in the individual
monograph. For example, for Phytonadione Injection, USP,
the pyrogen test dose is 2 ml per kg, while Protamine Sulfate
134 Chapter 2
Injection, USP, requires only 0.5 ml per kg containing 10 mg
per ml. Some injectable monographs specify the pyrogen test
dose on a weight-weight basis; for example, the dose of Diaze-
pam Injection, USP, is 0.25 mg per kg.
The test solution must be warmed to 37C prior to injec-
tion. The ear vein is swabbed with alcohol (70%), which not
only disinfects, but also improves visibility of the vein. Vein
longevity can be preserved by employing correct technique in
making the injection. A suggested procedure is as follows:
1. Rest the ear against the ngers of the left hand and hold
the ear down with the thumb (see Fig. 2.8).
2. Introduce the needle with the bevel edge upward near the
tip of the ear vein.
3. Slowly inject a small amount of sample to determine if the
needle is within the vein lumen. If not, a bubble will form
or backpressure will be felt. Withdrawing the needle
Fig. 2.8 Injection of pyrogen test sample into ear vein of rabbit.
Pyrogen Testing 135
slightly and moving it forward again should place it in
proper position.
4. Maintain steady pressure on the syringe plunger and com-
plete the injection within 10 minutes. Usually, the time
duration for infusion is much less than 10 minutes.
5. Withdraw the needle and apply pressure with the thumb
at the site of injection to retard bleeding and scarring.
Rectal temperatures are recorded at 1, 2, and 3 hours sub-
sequent to the injection. During the test period, rabbits and
equipment should be checked periodically. Occasionally, a rab-
bit may experience rectal bleeding, irritation, or leg or back
discomfort. Thermocouple wires might break or the electronic
thermal recorders may malfunction. Immediate action should
be taken in any of these situations.
Mazur and McKendrick (23) reported on the automated
pyrogen test systemused by McGawLaboratories. The system
manages all phases of the pyrogen test, including setting up
the test, acquiring and recording animal temperature data,
calculating test results, and issuing release reports. Today,
most modern pyrogen testing laboratories utilize similar com-
puter technology.
Test InterpretationUSP
According to the July/August 1991 issue of Pharmacopeial Fo-
rum, the solution may be judged nonpyrogenic if no single rab-
bit shows a rise in temperature of 0.5C or greater above its
control temperature. If this condition is not met, the test must
proceed to a second stage. There is no longer a second condi-
tion involving the sum of individual temperatures. In the sec-
ond stage, ve additional rabbits are given a new preparation
of the same test sample as the original three rabbits. The solu-
tion may be judged nonpyrogenic if not more than three of the
eight rabbits show individual temperature rises of 0.5C or
more.
The U.S. Public Health Requirements for Biological Prod-
ucts, Part 73, judge a solution to be pyrogenic if at least half
136 Chapter 2
Table 2.3 Comparison of United States Pharmacopeial (USP)
and British Pharmacopoeial (BP) Pyrogen Tests Requirements
Maximum total Minimum total
peak response (C) peak response (C)
to pass the test to fail the test
Number of
rabbits USP BP USP BP
3 1.4 1.15 1.4 2.65
6 2.80 4.30
8 3.7 3.7
9 4.45 5.95
12 6.60 6.60
of the rabbits tested show a temperature rise of 0.6C or more,
or if the average temperature rise of all rabbits is 0.5C or
more.
The British Pharmacopoeia (BP) (24) pyrogen test em-
ploys a sliding scale based on 3 rabbits and additional groups
of 3 rabbits, if required, for a total of 12 rabbits. This scale
is shown in Table 2.3 with the former USP test included for
comparison.
Limitations of the USP Rabbit Pyrogen Test
The USP rabbit pyrogen test suffers from several limitations,
which established the opportunity for the Limulus Amebocyte
Lysate test as a possible alternative for the rabbit test as an
ofcial pyrogen test procedure.
In Vivo Model
A test method (the in vivo model) that uses a living animal as
its model certainly must submit to a number of problems of-
fered by biological systems. Variability in biological systems
poses a great problem. No two rabbits will possess exactly the
same body temperature or respond identically to the same py-
rogenic sample. Rabbits are extremely sensitive and vulnera-
ble to their environment. This translates into an expensive
Pyrogen Testing 137
proposition in terms of facilities, control of the environment,
and training of the animal.
Pyrogen testing of rabbits is not only expensive, but also
laborious. Several hours are consumed in performing the pyro-
gen test, including a great amount of preliminary effort in pre-
paring the animals. Rabbits must be fed and watered properly,
cages cleaned to prevent disease, and time spent in training
the animals to adapt to the conditions of the pyrogen testing
facility and the test itself.
Rabbit Sensitivity to Pyrogens
The pyrogenic response in rabbits is dose dependent. The
greater the amount of pyrogen injected per kilogram body
weight, the greater the temperature increase in rabbits. This
is demonstrated in Table 2.4, taken from a report by Mascoli
and Weary (25).
A collaborative study initiated under the auspices of the
Health Industry Manufacturers Association (HIMA) demon-
strated that rabbits from 12 laboratories consistently failed
the (pyrogenic) test at doses of 1.0 ng per ml (10 ml/kg of 10
ng/kg endotoxin) of Escherichia coli 055: B5 endotoxin, and all
colonies passed (no pyrogenicity) at the 0.156 ng/kg dose (or
0.156 ng/ml using a 10 ml/kg dose) (25). The same study re-
ported that the average rabbit colony will attain a 50% pass/
fail rate with 95%condence at an endotoxin level above 0.098
ng/ml (10 ml/kg dose). The LAL test generally will detect en-
dotoxin levels of 0.025 ng/ml or less. Thus, the rabbit test is
less sensitive to endotoxin than the LAL test is.
Rabbit-to-rabbit variation in response to the same lot of
pyrogenic solution was shown by Mascoli and Weary (25). As
seen in Table 2.4, the standard deviations and coefcient of
variation values are rather high among eight rabbits adminis-
tered identical doses of endotoxin. The HIMA study reported
that, of 12 laboratories conducting rabbit pyrogen tests, 4
passed a level of 2.5 ng endotoxin per kg (26).
Sensitivity of the rabbit bioassay for endotoxin appears
to fall in the range of 1 to 10 ng/kg (19,27). Greisman and
1
3
8
C
h
a
p
t
e
r
2
Table 2.4 Eight Rabbit Pyrogen Test Results in Saline with Escherichia coli 055: BS Using
35 kg Rabbits
USP total Mean
E. coli endotoxin Volume solution temperature temperature Standard Coefcient of
concentration (ng/ml) injected (ml/kg) increase (C) increase (C)
a
deviation (C)
b
variation (%)
3.125 1.0 7.80
c
0.975 0.246 25.2
1.56 1.0 4.75
c
0.594 0.218 36.7
1.00 1.0 3.70
c
0.462 0.158 34.2
0.78 1.0 1.40 0.144 0.208 144.4
0.39 1.0 1.00 0.088 0.187 212.5
0.195 1.0 1.20 0.150 0.065 43.3
a
Negative rabbit temperature values were excluded from total temperature increase determinations
according to USP.
b
Negative rabbit temperature values were included in the determinations of means and standard devi-
ations to properly reect total variability.
c
Failed USP test criteria of 3.7C total increase.
Source: Ref. 25.
Pyrogen Testing 139
Hornick (19) found that the threshold pyrogenic dose of E. coli
endotoxin for both rabbits and humans is 1.0 ng/kg of body
weight. This holds true regardless of the volume of pyrogenic
solution administered because of the dose (rather than concen-
tration) dependency of the rabbit response to pyrogen.
Rabbit sensitivity to endotoxin varies with the time of day
(circadian) and time of year (cirannual) (28). The greatest rise
in temperature for any given dose of endotoxin occurred in the
afternoon, while the least rise occurred at midnight. At mid-
night, the greatest sensitivity was seen at the end of October,
while the least was seen at the end of April. However, this
was opposite at 10:00 A.M. Although not practical at all, it was
suggested in this report that a rabbit colony be tested for its
threshold sensitivity at the beginning of each month and at
the hours when products would be tested normally. Thus, sea-
sonal variability in sensitivity may be controlled.
Interferences of the Rabbit Pyrogen Test
Many products administered parentally cannot be tested for
pyrogens with the rabbit test because of the interferences they
create in the rabbit response to pyrogens if they are present
in the product. Any product having a pyretic side effect, such
as the prostaglandins and the cancer chemotherapeutic
agents, will interfere with the rabbit response. Several prod-
ucts are inherently toxic to the rabbit (see Table 2.5) and must
Table 2.5 Examples of Drugs and Drug Products Not Suitable
for Testing by the USP Pyrogen Test
Most cancer chemotherapeutic agents
Most anesthetics, muscle relaxants, and sedatives
Sterile betamethasone sodium phosphate solution
Chlorpheniramine injection
Magnesium sulfate
Metocurine iodide injection
Perphenazine
Thiopental sodium for injection
140 Chapter 2
be diluted to concentrations far below the pharmacologically
effective dose of the drug.
Despite these major limitations and the insurgence today
of the LAL test, it must not be forgotten that the USP rabbit
pyrogen test for decades has nobly served as a sufciently sen-
sitive test for pyrogens and has helped to eliminate pyrogenic
contamination fromdrugs reaching the marketplace, although
most pharmaceutical and device manufacturers currently use
the LAL test for the pyrogen test.
The ofcial referee test according to the USP and EP is
the LAL gel clot test.
THE LIMULUS AMEBOCYTE LYSATE TEST
History and Background
Credit for discovering the interaction between endotoxin and
the amebocyte lysate of the horseshoe crab, Limulus poly-
phemus, belongs to Levin and Bang (29). Basing their work
on earlier research by Bang (30), these workers were involved
in the study of clotting mechanisms of the blood of lobsters,
sh, and crabs. Autopsies of dead horseshoe crabs revealed
intravascular coagulation. The clotted blood was cultured and
found to contain gram-negative bacteria such as E. coli and
Pseudomonas. Further tests showed that amebocyte cells of
the blood of the horseshoe crab blood were extremely sensitive
to the presence of endotoxin, the toxic substance liberated by
the disintegration of bacterial cells. The substance in the
amebocytes responsible for reacting with endotoxin is known
to be a clottable protein, discussed in the section, Reaction
Mechanism. In lysing the amebocyte cells by osmotic effects,
a most sensitive biochemical indicator of the presence of endo-
toxin was produced, hence the name Limulus Amebocyte Ly-
sate test.
Limulus polyphemus (see Fig. 2.9) is found only at specic
locations along the East Coast of North America and the
coasts along Southeast Asia. The hearts of mature crabs are
punctured and bled to collect the circulating amebocyte blood
Pyrogen Testing 141
Fig. 2.9 Limulus polyphemus, the source of Limulus amebocyte
lysate reagent.
cells. Carefully performed, this procedure is not fatal to the
crab, and on proper restoration, the crab can be used again.
Since amebocytes act as activators of the coagulation mecha-
nism in the crab, an antiaggregating agent must be added to
inhibit aggregation. N-Ethylmaleimade is the most commonly
used anti-aggregant.
Amebocyte cells are collected and washed by centrifuga-
tion and lysed using distilled water. Lysing can also be done
with ultrasound, freezing and thawing, and grinding in a glass
tissue homogenizer (31). After lysing, the suspension is
142 Chapter 2
cleared of debris by centrifugation, and the supernate is lyoph-
ilized. Lyophilization is necessary for stability purposes. LAL
reagent is extremely sensitive to heat, and even in the lyophi-
lized state must be stored in the freezer (32). On reconstitu-
tion, LAL has a shelf life of 1 month storage at freezing condi-
tions.
The LAL test for pyrogens in parenterals was rst ap-
plied by Cooper et al. (33). The LAL test was found to be more
sensitive and, certainly, more expedient than the rabbit pyro-
gen test in the testing of radioactive drug products. Mallinc-
krodt, Incorporated, established the rst successful, large-
scale production facility for LAL in Chincoteague, Virginia, in
1971 (34).
On January 12, 1973 (Federal Register 38, 1404), the
Food and Drug Administration (FDA) stated that LAL was a
biological product and thus was subject to licensing under Sec-
tion 351 of the Public Health Service Act. Specications con-
cerning the purity and potency of LAL were proposed by the
FDA Bureau of Biologics (now CBER) later that year (Septem-
ber 18, 1973; 38 FR 26130). In the ensuing years, available
data on and experience with the LAL test accumulated, with
the primary use of the test being an in-process endotoxin test.
Finally, the FDA announced conditions under which the LAL
test could be used as an end-product test for licensed biological
products and medical devices (November 4, 1977; 42 FR
57749). This was followed by a draft guideline published by
the Ofce of Medical Devices for using the LALtest for medical
devices exclusively (March 20, 1979).
In the Federal Register of January 18, 1980 (45 FR 3668),
the FDA published a notice announcing the availability of a
draft guideline describing the conditions for validating the
LAL test before using it as a nal end-product endotoxin test
for human and veterinary injectable drug products. Com-
ments on the two draft guidelines (March 1979 and January
1980) resulted in a single draft guideline for validation of the
LAL test as an end-product endotoxin test for human and ani-
mal parenteral drugs, biological products, and medical de-
Pyrogen Testing 143
vices; this was published on February 2, 1983, and announced
on March 29, 1983 (48 FR 13096).* Specic details of this
guideline are identied in later sections of this chapter.
Until 1977, the Bureau of Biologics prepared its own ly-
sate. Since then, the Bureau has found it more economical to
purchase licensed lysate fromone of several licensed manufac-
turers. The specications required by the FDA before purchas-
ing a lot of LAL are summarized in Table 2.6 (36).
Reaction Mechanism
Elucidation of the endotoxin-LAL reaction has resulted pri-
marily from the work by Liu et al. (37), Takagi et al. (38), and
Mosesson et al. (39). Combining the results of these research-
ers efforts produces the following proposed reaction:
1. Endotoxin or a suitably prepared lipid A derivative of en-
dotoxin activates a proenzyme of LAL having a molecular
weight of 150,000.
2. Activation also depends on the presence of divalent metal
cations such as calcium, manganese, or magnesium. It has
been shown that the sensitivity of the LAL assay for endo-
toxin detection can be increased 10 to 30 times by using
LAL reagent containing 50 mM magnesium (40).
3. The activated proenzyme, related to the serine protease
class containing such enzymes as thrombin, trypsin, and
factor Xa, subsequently reacts with a lower molecular
weight protein fraction (MW 19,00025,000) contained
also in the LAL substance.
4. The lower molecular weight fraction, called coagulogen,
is cleaved by the proenzyme into a soluble and insoluble
subunit. The insoluble subunit appears as a solid clot, a
precipitate, or a turbid solution, depending on the amount
of insoluble coagulogen by-product formed.
* In December 1987, the FDA published its nal guideline on validation of the LAL
test as an end-product test for endotoxin for all products (human and animal paren-
teral products, biological products, and medical devices) (35).
144 Chapter 2
Table 2.6 Summary of FDA Standards Governing
the Manufacture of Limulus Amebocyte Lysate Reagent
Use of U.S. Standard Endotoxin for determining the sensitivity of
LAL.
Use of U.S. Reference LAL for establishing the potency of LAL.
Calculation of potency of each lot of LAL and the U.S. Reference
LAL using the U.S. Standard Endotoxin.
a. Test a minimum of 20 to a maximum of 28 vials per each dry-
ing chamber.
b. The 99% ducial upper limit of the standard deviation of the
log ratio of reference and test lysates for 20 vials can be no
greater than 0.73.
General requirements.
a. Handle horseshoe crabs in a manner to enable them to be re-
turned alive to their natural environment after a single collec-
tion of blood.
b. Perform sterility test on bulk lot and on each lling.
c. Run negative control tests of lysate.
d. Test for residual moisture.
Various labeling requirements.
Appropriate number of samples (not fewer than 28 vials) and docu-
mentation of manufacture of each lling, dates of testing, and
results of all tests must be submitted to Director, Bureau of
Biologics, FDA.
Source: Ref. 36.
Therefore, the coagulation reaction requires three factors
in addition to endotoxin. These three factorsa clotting en-
zyme, clottable protein (coagulogen), and certain divalent ca-
tionsare found in the LAL reagent. A schematic representa-
tion of the LAL reaction mechanism is found in Fig. 2.10
(41).
LAL Test Procedure
Cooper (42) rst described the methods and materials re-
quired to perform correctly the LAL test for pyrogen. While
the LAL test is a relatively simple procedure, especially when
Pyrogen Testing 145
Fig. 2.10 Schematic representation of the LAL reaction mecha-
nism. (From Ref. 41).
compared with the USP rabbit test, certain specic conditions
must be met. These include:
1. All materials that will come into contact with the LAL re-
agent or test sample must be thoroughly cleaned and de-
pyrogenated.
2. The reaction temperature cannot be outside the range 36
38C.
3. The reaction mixture must be within the range of pH 5
7.
4. The reaction time should be no longer than 1 hour.
5. Each test must be accompanied by positive and negative
controls.
The basic procedure of the LAL test is the combination
of 0.1 ml test sample with 0.1 ml LAL reagent. After 1 hour
incubation at 37C, the mixture is analyzed for the presence
of a gel clot. The LAL test is positive, indicating the presence
of endotoxin, if the gel clot maintains its integrity after slow
inversion of the test tube containing the mixture (see Fig. 2.11).
Complete instructions for conducting the LAL test are
found in inserts supplied with LAL test kits from commercial
manufacturers. The USP (22nd edition) also contains instruc-
tions for using the LAL test to estimate the concentration of
146 Chapter 2
Fig. 2.11 Apositive LALgel clot test is characterized by the forma-
tion of a solid gel that remains intact in the bottom of the tube on
inversion. (Courtesy of BioWhittaker, Inc., Walkersville, Maryland).
bacterial endotoxins in sample materials. These instructions
are summarized with commentary below:
Preliminary
1. Strict aseptic technique must be used to avoid microbial
contamination while conducting the test.
2. All containers and equipment used must be pyrogen free.
Heating at 250C or above for at least 60 minutes should
depyrogenate these items.
3. All glassware should be washed with detergent prior to
dry heat depyrogenation. If detergent is not completely
Pyrogen Testing 147
rinsed, it will interfere with the reaction and cause a false-
negative result.
4. Abide by all precautions in reconstituting and storing the
test reagents. Do not store diluted endotoxin used to deter-
mine LAL sensitivity because of loss of activity by adsorp-
tion to glass surfaces. The normal shelf life for LAL
reagent is 4 weeks at freezing temperatures after reconsti-
tution.
Standards
For drugs, biological products, and medical devices, the endo-
toxin standard is called the U.S. Standard Endotoxin or the
USP Reference Standard Endotoxin (RSE). The rst RSE lot
was designated as Lot EC-2 and had a dened activity of 1
endotoxin unit (EU)* in 0.2 ng of the standard (43). The cur-
rent FDA and USP reference standard is puried lipopolysac-
charide from E. coli 0113. One vial contains 10,000 EU.
When the USP selected the FDA endotoxin standard (pu-
ried lipopolysaccharide from E. coli 0113) as the new USP
reference standard (with established potency in endotoxin
units), this gave manufacturers the opportunity to standard-
ize their own control standard endotoxin (CSE) against the
USP RSE.
There are three LAL manufacturers licensed by the U.S.
government (44). Each manufacturer must determine the sen-
sitivity of each lot by using the U.S. Standard Endotoxin EC-
5, which is identical to the USP Endotoxin Standard (Lot F).
The FDA tests each lot for potency before releasing it to be
marketed.
* It has become accepted practice to use endotoxin units as the more desirable expres-
sion of endotoxin strength than weight or concentration terms. The use of endotoxin
units will allow any endotoxin type or lot to be used as a reference lot because its
activity can always be related to the original U.S. Reference Standard lot. This chap-
ter uses the endotoxin unit term as much as possible, but most literature references
cited use the weight or concentration terms as reported in the published articles. It
is noted in USP XXV and the 2002 EP that 1 EU 1 IU (international unit).
148 Chapter 2
If a manufacturer chooses to use an endotoxin prepara-
tion (CSE) other than the U.S. RSE, the CSE will have to be
standardized against the RSE. What this means is that the
CSE reaction in the rabbit, its uniformity, its stability, and its
interaction to a particular LAL lot all must be determined and
related to these same characteristics of the RSE.
1. At least four vials of the lot of CSE should be assayed by
determining end points (gelations) with LAL. The values
obtained should be the geometric mean of the end points
using a minimum of four replicates.
2. The end point for the CSE is stated in nanograms per mil-
liliter. The end point for the RSE is endotoxin units per
milliliter. So, if the LAL end point for the CSE is 0.018
ng/ml and the LAL end point for the RSE is 0.3 EU/ml,
then
RSE 0.3 EU/ml 16.6 EU/ng of CSE
CSE 0.018 ng/ml
3. This indicates that 0.018 ng of the CSE is equal to 0.3 EU
of the RSE. Thus, the CSE contains 16.6 EU/ng.
Validation of the LAL Test
To validate the use of the LAL test for any application requires
two determinations: initial qualication of the laboratory and
inhibition or enhancement properties of the product on the
LAL-endotoxin interaction. Extensive details of LAL test vali-
dation requirements are found in the Guideline on Validation
of the Limulus Amebocyte Lysate Test as an End-Product En-
dotoxin Test for Human and Animal Parenteral Drugs, Biolog-
ical Products, and Medical Devices (35).
Qualication of the laboratory simply involves using the
selected test method (gel clot end point, chromogenic and end
pointturbidimetric, or kinetic-turbidimetric techniques) to
determine its variability, to test new lots of lysate before use,
and to qualify the ability of the analyst(s) to conduct the test.
The LAL reagent used must have a conrmed potency (sensi-
Pyrogen Testing 149
tivity). This is achieved by combining the particular reagent
with a series of concentrations of RSE or CSE endotoxin,
bracketing the stated sensitivity (EU/ml) of the LAL reagent.
Use four replicates per concentration of endotoxin. The series
of endotoxin concentrations is prepared by twofold dilutions
of the RSE or CSE endotoxin using LAL-negative water for
injection. Following incubation and end-point determination
(manual or instrumental), the sensitivity of the LAL reagent
will be conrmed if the test results are positive to within one
twofold dilution of the stated label potency.
Inhibition/enhancement testing must be performed on
undiluted drug products or diluted drug products not ex-
ceeding the maximum valid dilution value (see Table 2.7) (35).
At least three production batches of each nished product
should be tested. The product is spiked with various known
amounts of RSE (or CSE), bracketing the sensitivity of the ly-
sate used, using four replicate reaction tubes per level of endo-
toxin. The same number of tubes is used for drug product con-
taining no added endotoxin and for control water for injection
samples also spiked with various known amounts of RSE or
CSE. The LAL test procedure is carried out manually or in-
strumentally.* The end points (E in units per ml) are then
observed and recorded for all replicate samples.
The end points are determined followed by computation
of the geometric mean of these end points. Geometric mean is
E (end points)
f (number of replicates)
and this mean is calculated for the control and test samples.
An illustration is given in Table 2.8 (41). The geometric means
of the product sample and the water control sample are com-
pared. If the product sample mean is within twofold of the con-
* The FDA validation guideline contains specic directions for inhibition/enhance-
ment testing depending on the technique usedgel clot, inorganic and end point
turbidimetric, and kinetic turbidimetric.
150 Chapter 2
Table 2.7 Examples of Minimum Valid Concentration (MVC)
and Minimum Valid Dilution (MVD) Calculations
MVC determination
MVC M/K
Sensitivity of LAL reagent in endotoxin units per milliliter
M Rabbit dose or maximum human dose per kilogram
K 5.0 E/kg (0.2 EU/kg for intrathecal drugs)
If LAL sensitivity () was 0.065 EU/ml and the maximum hu-
man dose were 25 mg/kg, then the MVC would be
MVC
0.065EU/ml 25 mg/kg
5.0 EU/kg
0.325 mg/ml
If this dose were to be given intrathecally, the denominator
would be 0.2 EU/kg.
MVD determination
MVD
Potency of product
MVC
1: 61.5
If the potency of a product were 20 mg/ml, the MVD would be
MVD
20 mg/ml
0.325 mg/ml
1: 61.5
Therefore, this product can be diluted to 61.5 times its original
volume and still be able to detect the lower endotoxin concen-
tration limit by the LAL test.
Source: Ref. 35.
trol mean sample, the drug product is judged not to inhibit
or enhance the LAL-endotoxin reaction. For example, if the
product sample showed a geometric mean of 0.4 EU/ml and
the water control mean was 0.2 EU/ml, the LAL test is valid
for that product.
If endotoxin is detectable in the untreated specimens un-
der the conditions of the test, the product is unsuitable for the
inhibition/enhancement test. Either endotoxin must be re-
moved by ultraltration or further dilution can be made as
Pyrogen Testing 151
Table 2.8 Example of Geometric Mean Determination for a
Small-Volume Parenteral Product Undergoing LAL Testing for
Endotoxin
a
Gel end-point
results for
specimen dilutions
Replicates Unit End-point dilution
(f) y 0.5 0.25 0.125 factors (E)
1 0.25
2 0.5
3 0.5
4 0.25
5 0.5
E 2.0
a
Geometric mean E/f 2.0/5 0.4.
Source: Ref. 41.
long as the minimum valid dilution (MVD) is not exceeded,
and the inhibition/enhancement test repeated. If the drug
product is found to cause inhibition or enhancement of the
LAL test, the following courses of action can be taken (35):
1. If the drug product is amenable to rabbit testing, then the
rabbit test will still be the appropriate pyrogen test for
that drug.
2. If the interfering substances can be neutralized without
affecting the sensitivity of the test or if the LAL test is
more sensitive than the rabbit pyrogen test, then the LAL
test can still be used.
3. For those drugs not amenable to rabbit pyrogen testing,
the manufacturer should demonstrate that the LAL test
can detect the endotoxin limit established for the particu-
lar drug. If the limit cannot be met, the smallest quantity
of endotoxin that can be detected must be determined.
There are various miscellaneous requirements in the pro-
cedures for validating the LAL test:
152 Chapter 2
1. Use positive and negative controls in all tests.
2. Use the highest and lowest drug concentrations for drug
products marketed in three or more concentrations.
3. Use three lots of each drug concentration for the valida-
tion tests.
4. If the lysate manufacturer is changed, the validation test
must be repeated on at least one unit of product.
5. The LAL reagent should have a sensitivity of at least 0.25
EU/ml.
6. The endotoxin control must always be referenced to the
RSE.
7. Any change in the product formulation, manufacturing
process, source of formulation ingredients, or lot of lysate
necessitates a revalidation of the LAL test for the product.
The possibility of a device inhibiting or enhancing the
LAL-endotoxin reaction is determined by extraction testing of
each of three device production lots. The extract solution must
be pyrogen-free water or saline to which known amounts of
standard endotoxin, bracketing the sensitivity of the lysate,
have been added. Depending on the type of device, extracts
may be obtained by ushing, immersing, or disassembling,
then immersing the device with the endotoxin-spiked solution.
The LALtest results of the extract should not be different from
the results of testing standard solutions containing endotoxin
that have not been exposed to the device.
Endotoxin highly adsorbs to container surfaces. Novitsky
et al. (45) reported on the different adsorptive natures of con-
tainer surfaces. Recovery of endotoxin occurred with polysty-
rene containers, while the worst for recovering endotoxin were
polypropylene containers. In fact, regardless of extraction
method, less than 1% endotoxin was ever recovered from poly-
propylene containers. Bonosilicate glass allowed higher recov-
ery than int glass.
Great care must be exercised in preparation and storage
of parenteral vials used for LAL testing. Guilfoyle et al. (46)
reported that 2040%of spiked endotoxin in vials was lost due
Pyrogen Testing 153
to adsorption to rubber stoppers. The authors suggested that
product containers be stored in an upright position and a uni-
form mixing procedure prior to assay be established.
Manual LAL Test Procedure
Four or more replicate samples at each level of the dilution
series for the test samples are used in most cases. The pH of
the reaction mixture must be between 6.0 and 7.5 unless speci-
ed differently in the particular monograph. The pH may be
adjusted by addition of sterile, endotoxin-free 0.1 N sodium
hydroxide, 0.1 N hydrochloric acid, or suitable buffers.
Test tubes, usually 10 by 75 mm, are lled with an ali-
quot, usually 0.1 ml, of reconstituted LAL reagent and the
same aliquot volume of the test sample. In other test tubes,
equal volumes of LAL reagent and endotoxin standard are
combined. Positive controls (LAL reagent sample containing a
known concentration of endotoxin) and negative controls (LAL
reagent plus an equal volume of sterile, pyrogen-free solvent)
are run simultaneously with the test samples and endotoxin
standards.
When the equal volumes are combined, the test tube is
swirled gently. The tube is placed in a constant temperature
water bath with temperature controlled at 37C 1C. Incu-
bation times ideally last 60 2 minutes. While incubating,
the test tubes must never be disturbed for fear of irreversibly
disengaging the gel clot if it has formed. Careful removal of
the incubated test tubes for gel clot analysis is extremely im-
portant.
The use of microscope slides containing petrolatum wells
has been advocated for conducting the LAL test when lower
reagent consumption is desired (47). One slide can accommo-
date 12 samples using microliter volumes (0.1 l). A dye solu-
tion (0.1% toluidine blue in ethanol) is placed in each well to
aid in interpreting the results. A positive LAL test generates
a blue star in the droplet, while a negative LAL test gives
a homogeneous blue solution.
The degree of gel formation can be determined by either
154 Chapter 2
direct visual observation or instrumental analysis. Visual ob-
servation starts by carefully removing the test tube from the
incubator, then carefully inverting (by 180) the test tube and
visually checking for the appearance of a rm gel. A positive
reaction is characterized by the formation of a rm gel that
does not break or lose its integrity during and at the comple-
tion of the inversion process. Anegative result is characterized
by the absence of a gel or by formation of a viscous gel that
does not maintain its integrity during the inversion process.
An example of a positive LAL test result is seen in Fig. 2.11.
Instrumental Tests
Direct visual observation of the gel end point relies on the sub-
jective interpretation of the observer and, unless twofold serial
dilutions are performed, provides only a qualitative (yes or no)
measurement of the endotoxin present in the sample. Analysis
of the gel end point by instrumental methods offers several
advantages, including single-tube quantitation and objectiv-
ity. In addition, instrumental methods can be automated, re-
sulting in increased speed, efciency, and adaptation to com-
puter control.
Two basic instrumental methods are available for LAL
testing. One method is based on turbidimetric measurement
of gel formation (e.g., Abbotts MS-2, Millipores Pyrostat),
while the other method is based on colorimetrically measuring
a chromophobic substance produced during the LAL-endo-
toxin reaction (e.g., Mallinckrodt and BioWhittaker).
The Abbott MS-2 Microbiology Systemwas designed orig-
inally for automated antibiotic susceptibility testing of clinical
samples. The system was rst described by Jorgensen and Al-
exander (48) and later by Novitsky et al. (49). A general proce-
dure is outlined below:
1. LAL is mixed with the test sample in a 1: 4 ratio, (e.g., 100
l LAL 400 l test) in a polystyrene research cuvette.
2. Up to 88 samples can be incubated per module. Incubation
occurs at 35C for 60 minutes.
3. The mixture of each cuvette following incubation is exam-
Pyrogen Testing 155
ined for turbidity (light transmission) by recording the op-
tical density (OD) at 670 nm on the MS-2 spectrophotome-
ter. The samples are examined at either 1- or 5-minute
intervals.
4. The OD values are recorded on a cassette tape and/or pa-
per and can be displayed graphically as ODversus time on
a cathode ray tube or transferred to paper with a hardcopy
printer. An example of a plot of OD at 670 nm versus time
using standard endotoxin samples is shown in Fig. 2.12
(49).
Turbidimetric OD has been shown to be directly propor-
tional to E. coli endotoxin concentration on a log-log plot. For
example, Fig. 2.13 shows such a relationship. Standard curves
usually are linear only within a relatively small concentration
range, for example, 0.010.1 ng/ml (0.11.0 EU/ml). The es-
tablishment of standard curves for instrumental analyses of
the LAL-endotoxin reaction can be difcult. The availability
of standard endotoxin has improved the reproducibility of
standard curve determinations.
Fig. 2.12 Turbidimetric response of LAL with control standard en-
dotoxin diluted in sterile water for irrigation. A, 100 pg endotoxin/
nl; B, 2.5 pg/nl; C, 6.3 pg/nl; D, 1.6 pg/nl; E, 0.4 pg/nl. (From Ref.
49).
156 Chapter 2
Fig. 2.13 Log-log relationship between turbidimetric optical den-
sity and endotoxin concentration. (Courtesy of Millipore Corp., Bed-
ford, Massachusetts).
Another aspect of the kinetic LAL test that presents chal-
lenges for the parenteral manufacturer is CSE spike recovery.
According to USP and EP, the CSE spike recovery in product
must be 50200%. Findings by Zink-McCullough et al. have
shown that shown that CSE can be recovered in a battery of
products by dilution (154). A practical example of the use of
the Abbott MS-2 automated LAL test system in the detection
of bacteriuria was published by Jorgensen and Alexander (50).
The use of turbidimetry in automated LAL testing provided a
way of successfully analyzing endotoxin in blood (51). Auto-
mated microliter testing overcomes the inhibitory factors in
blood that mask the gelatin reaction using conventional LAL
test methodology.
A newer type of automated LAL test system is based on
the measurement of color intensity of the LAL gel end point.
Pyrogen Testing 157
This systemis called the Chromogenic LAL assay system(Fig.
2.14). The test sample is mixed with LAL reagent and incu-
bated at 37C for a period of time (usually 10 minutes). A sub-
strate solution containing a color-producing substance is then
mixed with the LAL test sample and incubated at 37C for an
additional 3 minutes. The reaction is stopped with 50% acetic
acid. The color absorbency of the sample mixture is deter-
mined spectrophotometrically at 405 nm. The more intense
the color, the greater the absorbance value measured. Endo-
Fig. 2.14 An example of automated LAL technology is the kinetic
chromogenic method, one instrument of which is shown here. (Cour-
tesy of BioWhittaker, Inc., Walkersville, Maryland).
158 Chapter 2
toxin concentration can then be determined from a standard
plot of absorbance versus endotoxin concentration in nano-
grams per milliliter or endotoxin units per milliliter.
The chemical composition of the substrate is a peptide
chain linked to p-nitroaniline (pNA) (52). The endotoxin cata-
lyzes the activation of a proenzyme in the LAL, as discussed
on pages 124125. The activated enzyme, in turn, catalyzes
the splitting of pNA from the colorless substrate. In Fig. 2.10,
pNA replaces coagulogen as the substance cleaved by the pro-
enzyme. It is pNA that is measured spectrophotometrically.
Absorbance at 405 nm and endotoxin concentration are lin-
early related between 0.01 and 0.1 ng/ml.
For laboratories responsible for conducting multiple LAL
tests, automation practically becomes a necessity. Automation
employs all the advantages of instrumental analyses, includ-
ing greater precision and sensitivity. Technology has ad-
vanced to the point at which the LAL test can be performed
automatically using robotic systems such as one produced by
Zymate (53). Such a system will automatically dilute a stock
reference endotoxin standard for construction of a ve-point
standard curve, make sample dilutions to the proper testing
concentration, and perform chromogenic substrate LAL
assays in duplicate. In 48 minutes, the automated system
assays three samples and a reference standard in duplicate
along with a water blank. The method can be sensitive to a
detection limit of 0.003 EU/ml with 30 minutes of incubation.
Assay precision is approximately 6%. The major disadvan-
tages of automated LAL testing systems are their cost and
complexity. Cooper (54) recommended that each laboratory
carefully consider its present and long-term needs and be
rmly grounded in the fundamentals of the LAL test before
changing from manual to automated LAL test systems.
Lindsay et al. (55) described a new reagent for the chro-
mogenic LAL assay. A single reagent now contains the LAL
components, buffer, and the chromogenic substrate.
The LAL test requirements for lack of pyrogenicity or
Pyrogen Testing 159
critical endotoxin concentration will be met if there is no for-
mation of a rm gel at the level of endotoxin specied in the
individual monograph. For instances when instrumental anal-
yses have been done, the sample will pass the LAL test if not
more than the maximum permissible amount of endotoxin
specied in the individual monograph is present in the sam-
ple. In addition, the condence limits of the assay must not
exceed the limits previously specied for the instrumental
analysis.
Endotoxin Limits in Parenteral Articles
Endotoxin limits are necessary because bacterial endotoxin is
ubiquitous and is expected to be present in all articles at some
level. The question is, What level is safe? This becomes the
endotoxin limit (57).
The rst FDA draft guideline for LAL testing of drugs
(58) proposed an endotoxin limit for all parenterals of 0.25 EU/
ml. This limit was vehemently opposed by the parenteral drug
industry because the limit was arbitrary, based on concentra-
tion rather than endotoxin quantity per dose, and did not per-
mit sufcient dilution of small-volume parenterals known to
inhibit the LAL test reaction.
The Parenteral Drug Association proposed an alternative
endotoxin limit based on rabbit or human dose (59); the FDA
accepted this alternative and it became part of the new FDA
draft guideline for end-product testing published in December
1987 (60). The new endotoxin limit is:
K Threshold Pyrogen Dose (TDP)
M Maximum rabbit or human dose
where the TPD has been dened as 5 EU/kg, the lower 95%
condence limit of the average dose found to produce a pyro-
genic response in rabbits and humans (61). For drugs adminis-
tered intrathecally, for which pyrogenic contamination can be
much more dangerous (see pp. 154155), the TPD is 0.2 EU/
kg.
160 Chapter 2
The maximumrabbit or human dose is that dose adminis-
tered per kilogram of body weight of rabbit or human in a pe-
riod of a single hour, whichever is larger. For example, if a
drug with a concentration of 1 mg/ml has a maximum human
loading of 25 mg/kg while the rabbit pyrogen test dose is 10
mg/kg, the maximum dose used in the denominator of the en-
dotoxin limit equation would be the human dose of 25 mg. On
the other hand, were the above human dose only 2.5 mg/kg,
then the rabbit dose of 10 mg/kg would be the larger of the
two doses. The endotoxin limit for the two examples would be
EU
5 EU/kg
25 mg/kg
0.2 EU/mg
EU
5 EU/kg
10 mg/kg
0.5 EU/mg
For devices, the endotoxin limit is 0.1 ng per milliliter of ex-
tract solution.
Four classes of drugs are exempted from the endotoxin
limit dened by K/M:
1. Compendial drugs for which other endotoxin limits have
been established
2. Drugs covered by new drug applications, antibiotic Form
5 and Form 6 applications, new animal drug applications,
and biological product licenses for which different limits
have been approved by the agency
3. Investigational drugs or biologics for which an IND or
INAD exemption has been led and approved
4. Drugs or biologics that cannot be tested by the LAL
method example
Schmitz (62) reviewed all the progress of the establish-
ment of endotoxin limits leading to the Guideline on Valida-
tion of the Limulus Amebocyte Lysate Test as an End-Product
Endotoxin Test for Human and Animal Parenteral Drugs, Bio-
logical Products, and Medical Devices (35). This guideline
contains a list of maximum doses per kilogram and the corre-
Pyrogen Testing 161
sponding endotoxin limits for a large number of aqueous in-
jectable drugs and biologics on the market. The fth supple-
ment of USP XXII subsequently listed 185 monographs with
newBacterial Endotoxin Test requirements based on the max-
imum recommended total dose.
Sensitivity of LAL
LAL sensitivity is dened as the lowest concentration of a puri-
ed endotoxin that will produce a rm gel that will remain
intact when inverted carefully after 1 hour of incubation at
37C. (LAL sensitivity is also expressed as how many times
its sensitivity is greater than the rabbit test.) In general, it
seems to be well established that the LAL test is sensitive to
picogram quantities of endotoxin, and that LAL is from 5 to
50 times more sensitive than the rabbit test to the presence
of endotoxin, depending on the type of comparative study con-
ducted.
Earlier studies by Cooper et al. (33) demonstrated that
the LAL test was at least ve times more sensitive to puried
endotoxin than the rabbit test. This was later conrmed by
Elin and Wolff (63). Improvements in LAL production and for-
mulation methodology increased the sensitivity of LAL to be
10 to 50 times greater than the rabbit test (42,64). These num-
bers were based on a gel time of 1 hour and a rabbit test dose
of 1 ml/kg.
Ronneberger (65) found that the LAL test gave similar
results or was 10 times more sensitive than the rabbit test
using lipopolysaccharides from different gram-negative bacte-
ria (see Table 2.9). In more than 300 samples of drugs, plasma
proteins, and other antigens, the LAL test and rabbit test gave
similar results, although injection of a higher volume of test
sample compensated for the lower sensitivity of the rabbit
test.
Marcus and Nelson (31) have stated that the rabbit pyro-
gen assay will detect 1 to 10 ng of enterobacterial endotoxin,
while the LAL test will detect 0.01 to 0.1 ng endotoxin per
162 Chapter 2
Table 2.9 LAL Specicity and Sensitivity for the Detection
of Lipopolysaccharides
Minimum
Minimum dose concentration
for positive for positive
LPS source rabbit response
a
LAL reaction
b
Salmonella typhi Type 58 1 ng 0.1 ng/0.1 ml
Salmonella abort. equi 10 ng 10 pg/0.1 ml
Lipid A of Salmonella 100 pg 10 pg/0.1 ml
abort. equi
Salmonella minnesot. 10 pg 1 pg/0.1 ml
E. coli 10 ng 0.1 ng/0.1 ml
Klebsiella pneumoniae 1 ng 2 ng no reaction
a
Three rabbits used.
b
LAL source: Pyrogent (Mallinckrodt).
milliliter of solution. The ability of LAL to detect E. coli endo-
toxin in pyrogen-free distilled water was found to be 100 times
more sensitive than the rabbit test (see Table 2.10) (66).
Lysate sensitivity will vary according to the commercial
source of the lysate, as is the case with endotoxin sensitivity.
Wachtel and Tsuji (67) tested six commercial lysate prepara-
tions against E. coli endotoxin. Sensitivity ranged from 0.03
to 0.003 ng/ml. Similar results were found with endotoxins
extracted from Salmonella typhosa, Serratia marcescens, and
Shigella exneri. Pseudomonas sensitivity ranged from 10 ng/
ml to as high as 500 ng/ml.
Twohy et al. (68) compared lysates from ve LAL manu-
facturers (Associates of Cape Cod, Difco, Haemachem, M. A.
Bioproducts, and Mallinckrodt). Using the gel clot method
with nine different drug products and EC-5 endotoxin stan-
dard, these investigators found that end points varied among
the LAL reagents from the different manufacturers. Lot-to-
lot variability with LAL reagent from two manufacturers was
Pyrogen Testing 163
Table 2.10 Sensitivity of the Rabbit Pyrogen Test and of the
Limulus Test in the Detection of E. coli Endotoxin
Rabbit
ng/ml pyrogen test
a
Limulus test
500 Pyrogenic Positive
50 Pyrogenic Positive
5 Nonpyrogenic Positive
0.5 Nonpyrogenic Positive
0.05 Negative
0.005 Negative
a
Dose: 1 ml endotoxin solution/kg body weight.
observed, as was the ability of some LAL reagents to change
the pH of the drug product. These data support the fact that
some LAL reagents are better suited for some drug products,
than for other products, and that the LAL test must be revali-
dated for every drug product when LAL manufacturers and/
or LAL lots are changed.
Sensitivity of LAL for endotoxin depends greatly on the
vehicle in which the endotoxin is contained. For example, LAL
can detect only 5 to 10 g/ml endotoxin in plasma, whereas
0.05 g/ml endotoxin was detectable in cerebrospinal uid
(69). The failure of LAL to detect known levels of endotoxin
in human serum albumin and other protein solutions is well
known (41). Many drug products inhibit the LAL test and se-
verely retard its sensitivity. These inhibitions and limitations
of the LAL test are discussed in the section, Limitations.
Test Specicity
Whereas sensitivity is the ability of a test to give positive reac-
tions in the presence of the material tested, specicity is the
ability of a test to give positive reactions with only the mate-
rial tested (31). The sensitivity of LAL toward endotoxin is
164 Chapter 2
undisputed. However, its specicity in reacting solely with en-
dotoxin is its most controversial characteristic.
In 1973, Elin and Wolff (63) rst reported the possible
lack of specicity of the LAL test for bacterial endotoxin. Sub-
stances found to cause lysate gelatin included thrombin,
thromboplastin, ribonucleases, and polynucleotides such as
polyriboadenylic acid and polyribouridylic acid. Wildfeuer et
al. (69) found that peptidoglycans isolated from various gram-
positive bacteria caused lysate gelatin. Positive reactions have
been found between LALand streptococcal exotoxins (70), syn-
thetic dextrans (71), lipoteichoic acids (72), and the dithiols
dithiothreitol and dithioerythritol (73). Intravenous immuno-
globulin treatment has been found to produce false-positive
LAL test results (74). Increasing the amounts of administered
immunoglobulins increased the levels of LAL-reactive mate-
rial in plasma.
Amibocyte lysate contains substances called (13) -D-
glucansensitive factors (75). These factors can activate LAL
to produce false-positive results for the presence of endotoxin.
Interestingly, very small amounts of -glucan (11000 ng/ml
plasma) will trigger gelation, while greater amounts of -glu-
can (1 mg/ml plasma) will not (76).
Pearson and Weary (77) addressed these false-positive re-
actions caused by nonendotoxin substances. They concluded
that such substances need not concern parenteral drug manu-
facturers because of one or more of the following reasons:
1. Many of the substances (including all of the synthetic sub-
stances) would not be found in a parenteral product.
2. The substance may be present, but not in sufcient con-
centrations to produce gelation in lysate or fever in rab-
bits.
3. The substance is a highly puried preparation that could
not occur in production.
4. Results have not been conrmed by other researchers.
5. Because a negative LAL test result demonstrates the un-
equivocal absence of endotoxin, concern over false posi-
Pyrogen Testing 165
tives becomes a moot point with proper positive and nega-
tive controls.
Concerns over false negatives can be eliminated by the same
validation process. For example, the clotting enzyme in LAL,
coagulogen, is similar biochemically to trypsin. Trypsin, in
turn, can initiate the gelation reaction (34). To be certain that
a positive LAL test is due unequivocally to endotoxin contami-
nation, adequate controls are used to demonstrate that sub-
stances like trypsin are not the cause of the gelation observed.
Advantages Compared to the USP Rabbit Test
Proponents of the LAL test claim that the test offers at least
seven advantages over the use of the USP rabbit test for de-
tecting pyrogens in parenteral injectable products and medi-
cal devices (25):
1. Greater sensitivity
2. Greater reliability
3. Better specicity
4. Less variation
5. Wider application
6. Use as a problem-solving tool
7. Less expense
The majority of these advantages are a direct result of the re-
markable simplicity of the LAL test. Being an in vitro test re-
quiring a minimal number of items to complete the test, LAL
offers rapidity and reliability unmatched by an in vivo system.
Control of technique, handling, and external environmen-
tal factors is achieved much more easily with the LAL test.
This, in turn, leads to minimized chances of error and varia-
tion in the testing results. The ease and adaptability of the
LAL test allow it to be used in many different situations for
which application of the rabbit test would be impractical or
impossible. In fact, the need for a rapid, simple, and sensitive
technique for pyrogen testing of extemporaneously prepared
radiopharmaceutical preparations led to some of the earliest
166 Chapter 2
applied research involving the LAL test in hospital pharmacy
quality control. Other applications of the LAL test, rendered
possible because of its unique advantages compared to the rab-
bit test, include pyrogen testing of in-process water for injec-
tion, bacterial and viral vaccines, antineoplastic agents, and
drugs designed for intrathecal injection and validation of dry
heat depyrogenation cycles. These applications are elaborated
in the section. LAL Test Applications.
The LAL test has become an acceptable substitute for the
rabbit test in the in-process pyrogen control of plasma frac-
tions (78). Four advantages given for substituting the LALtest
in place of the rabbit test were as follows:
1. An in vitro test, when available, is employed rather than
an animal test.
2. Results are available within 90 minutes after beginning
the test procedure.
3. Tests that can be conducted with LAL when using the rab-
bit test would be senseless because of the time factor.
4. The LAL test is simple and inexpensive.
Fumarola and Jirillio (79) stated that, according to some
140 papers reported in the literature dealing with the LAL
test as well as their own experience, the test is an acceptable,
specic, rapid, and sensitive method for endotoxin assay of
parenteral drugs and biological products and for in-process
testing of parenteral solutions.
Researchers at Travenol Laboratories have published
many articles providing data to support the superiority of the
LAL test over the rabbit test for pyrogen testing of LVPs
(25,3032). Their arguments were summarized by Mascoli
and Weary (25):
1. Pyrogens important in LVP products and devices are en-
dotoxin in nature.
2. After tens of thousands of tests, an unexplained negative
LAL test resultpositive rabbit test result was never re-
corded.
Pyrogen Testing 167
3. Some endotoxin pyrogens detected by LAL were detected
by rabbit tests.
4. In some cases, the rabbit test results only failed initially
to detect pyrogens that were sometimes conrmed later
by rabbit tests, but were always conrmed by initial LAL
tests.
In a poll taken by M. J. Akers of quality control represen-
tatives from 10 pharmaceutical manufacturers of parenteral
products and devices, 7 of 10 responded that they preferred
the LAL test over the rabbit test. The advantages of the LAL
test as reasons given for their preference were (in order of im-
portance)
1. Greater sensitivity
2. Less variation
3. Quantitative results
4. Less time consuming
5. Less expensive
6. An easier test
The reader is directed to a nationwide survey of the bio-
technology industry regarding practices also listed with endo-
toxin detection by the LAL assay (83).
Limitations of the LAL Test
Unquestionably, the LAL test lls the need for a simple, sensi-
tive, accurate, and inexpensive method for detecting bacterial
endotoxin. It certainly offers an excellent alternative or sup-
plemental method of the ofcial USP rabbit test for pyrogen.
However, it is not without limitations or problems.
The greatest limitation of the LAL test is the problem of
interference of the lysate-endotoxin interaction that is caused
by a variety of drugs and other substances (84,87). Of the 10
quality control representatives from the parenteral industry
we polled, 7 identied inhibition of the lysate-endotoxin inter-
action as the number one factor limiting the applicability of
the LAL test. As discussed on pages 124125, the LAL gela-
168 Chapter 2
tion reaction is mediated by a clotting enzyme that is heat
labile, pHsensitive, and chemically related to trypsin (8990).
Inhibition is caused by any material known to denature pro-
tein or to inhibit enzyme action. A representative listing of
drugs and other substances known to modify or inhibit the
lysate-endotoxin interaction is given in Table 2.11. Inhibition
by many drug components can be overcome by dilution or pH
adjustment. Of course, dilution reduces the concentration of
the endotoxin and places greater demand on the sensitivity of
the LAL reagent to detect diluted amounts of endotoxin.
Tests for inhibition or activation basically involve the use
of positive controls. Product samples are spiked with known
endotoxin levels, preferably the same levels used in standards
prepared for sensitivity determinations. The end point of de-
tection for the product sample should be no different from the
end point for the standards series. In other words, if the lowest
standard detectable level of endotoxin is 0.025 ng/ml, this
level must also be detectable by the same lot of LAL reagent
in the product sample. If inhibition is found to occur, serial
dilutions of the product sample are made until the appropriate
dilution is found that no longer modies the gelation reaction.
Inhibition
According to Cooper (90), 30% of drug products do not inhibit
the LAL test (producing an increase in the expected gelation
onset time). Of the majority of products that do inhibit the
test, 97% of the problems can be resolved because the inhibi-
tion is concentration dependent. Simple dilution usually can
LAL-endotoxin reaction.
LAL test inhibition is considered signicant if the posi-
tive control varies by more than a twofold dilution from the
standard in water. Inhibition acts on endotoxin, not the LAL
reagent; that is, inhibition is often a failure to recover inade-
quately dispersed liposaccharide (aggregation of puried en-
dotoxin).
Pyrogen Testing 169
Table 2.11 Examples of Small Volume Parenterals Reported
to Markedly Inhibit the LAL Test
Inhibition overcome by more Inhibition at maximum
than one twofold dilution valid dilution
Aminophylline injection Carbazochrome salicylate
Ascorbic acid and vitamin B Cyclizine lactate
complex injection
Chorionic gonadotropin Diatrizoate meglumine and dia-
trizoate sodium
Clindamycin phosphate Edetate disodium injection
Cyanocobalamin injection Fluorescein sodium
Dicyclomine hydrochloride Liver injection
Diphenydramine hydrochloride Meperidine hydrochloride and
promethazine hydrochloride
Dyphylline injection Oxacillin sodium and other pen-
icillin products
Ephedrine hydrochloride Pentamidine isethionate
Fluorouracil Peptonized iron large volume
parenteral
Lidocaine hydrochloride Sulsoxazole
Lidocaine hydrochloride and epi- Sulfobromophthalein sodium
nephrine
Meperidine hydrochloride in- Vancomycin hydrochloride
jection
Mepivacaine hydrochloride and Multi-vitamin injection
levonordefrin
Promethazine hydrochloride
Scopolamine hydrobromide
Tetracaine hydrochloride
Thiamine hydrochloride
Source: C. W. Twohy, A. P. Duran, and T. E. Munson, J. Parenter. Sci. Tech.,
38, 190201, (1984).
170 Chapter 2
Primary ways in which drug products inhibit the LAL
test are by
1. Suboptimal pH
2. Aggregation or adsorption of control endotoxin spikes
3. Unsuitable cation concentrations
4. Enzyme or protein modication
5. Non-specic LAL activation
Other concerns or limitations of the LAL test are as fol-
lows:
1. LAL is dependable only for the detection of pyrogen origi-
nating from gram-negative bacteria.
2. Being an in vitro test, the LAL test cannot measure the
fever-producing potential of endotoxin present in the
sample.
3. The sensitivity of LAL varies appreciably with endotoxins
from various microbial sources.
4. It is difcult to compare the sensitivity of the LAL test and
the rabbit test because the rabbit assay is dose dependent,
while the LAL test is concentration dependent.
5. Gel formation can be difcult to interpret and can be bro-
ken on the slightest vibration.
6. The LAL test is too sensitive in that it can detect endo-
toxin at levels below those required for producing fever in
mammals. Yet, the FDA may enforce a level of sensitivity
for the LAL test much greater than that for the rabbit test.
In other words, a product that will consistently pass the
USP pyrogen test may not pass the LAL test. Does this
mean that the product is pyrogenic and harmful to hu-
mans?
7. Potential interferences from -glucans.
8. Extensive studies are required to validate the LAL test as
the nal product pyrogen test.
Test Variability
There are several sources of variability that can affect the ac-
curacy and reliability of the LAL test. It is for these reasons
Pyrogen Testing 171
that validation is so important and why the FDA produced its
validation guideline for the LAL test (35). Pearson (15) and
McCullough (91) have written excellent reviews on this
problem.
1. Reagent Variability. There are signicant differences
in LAL reagent formulation from manufacturer to manufac-
turer (92). Although all LAL reagents are standardized to the
USP RSE, both manufacturing processes and formulation dif-
ferences account for variations seen in real-world endotoxin
test situations. Major differences in reagent preparation in-
clude addition of the following: divalent cations, albumin, buff-
ers, and surface-active agents. Some manufacturers allow the
crude reagent to age, adjust coagulogen concentration, and
perform chloroform extraction to remove inhibitors and in-
crease sensitivity.
2. Method Variability. LAL reagents are designed spe-
cically for optimal activity in each of the major LAL test sys-
tems. Thus, lysatedrug product compatibility may change
when switching from one test method to another using the
same lysate manufacturer.
3. Product Variability. It is well known that many paren-
teral products will interfere with the lysate-endotoxin reac-
tion, although most of these interferences can be overcome by
dilution (93).
4. Laboratory Variability. Type of glass and/or plas-
ticware used (94), equipment calibration procedures, recali-
bration procedures, purity of water used, dilution procedures,
and other different laboratory procedures all contribute to
LAL test variability. As discussed previously, differences in
handling (degree of agitation) and storage of parenteral prod-
ucts prior to LAL test analysis can markedly affect test re-
sults.
As a reiteration, to control all these sources of variability,
the FDA wrote its guideline on validation of the LAL test (35).
The guideline says, The USP inhibition/enhancement tests
must be repeated on one unit of the product if the lysate manu-
facturer is changed. When the lysate lot is changed, the two
172 Chapter 2
lambda positive control is used to re-verify the validity of the
LAL test for the product.
For an LAL reagent to be compatible with the FDA guide-
lines for LAL evaluation of drugs, devices, and biologicals and
with the USP Bacterial Endotoxin Test, the reagent should
have a stabilized sensitivity of 0.12 EU/ml. This sensitivity
should be referenced to an E. colidelivered LPS such as the
USP RSE from E. coli. An LAL reagent should be buffered to
accommodate small changes in pH of the test solution and be
stabilized for divalent cations. The reagent also should be spe-
cic for endotoxin and should exhibit a clear and accurate end
point.
LAL Test Applications
From a modest beginning of detecting endotoxin in blood, LAL
test application has expanded into a variety of laboratory and
clinical situations. New or improved usage of the LAL test ap-
pears in the literature on a monthly basis. Methodology has
become more standardized, reference standards more ac-
cepted, and automatic instrumental analysis has been devel-
oped. LAL testing for endotoxin in the parenteral eld has be-
come standard practice.
At this time, the LAL test has been used as an indicator
of endotoxin contamination in at least six different areas:
1. Pharmaceuticals
2. Biologics
3. Devices
4. Disease states
5. Food
6. Validation of dry heat cycles
The literature is massive with regard to LAL test applica-
tions in most of these areas. Not all published reports are dis-
cussed, but those with signicant impact are described in the
following sections. Reference 95 is a good source of articles
dealing with applications of the LAL test.
Pyrogen Testing 173
Pharmaceuticals
The LAL test has overtaken the rabbit test as the main nal
pharmaceutical product release test for pyrogens. More than
200 USP monographs now contain endotoxin limits using the
LAL test.
Radiopharmaceuticals represent a special class of paren-
teral medications for which the LAL test offers unique advan-
tages in the detection of pyrogen contamination. Many radio-
pharmaceuticals are prepared extemporaneously, such as
technetium 99m (
99m
Tc), which has a biological half-life of only
6 to 7 hours. The LAL test, because of its short time for testing,
low volume requirements, and low cost, obviously is the pre-
ferred method for pyrogen detection in radiopharmaceuticals.
DeMurphy and Aneiros (96) used a micro-LAL test method
for pyrogen detection in 204 radiopharmaceuticals, including
pertechnetate, sulfur colloid, pyrophosphates, pyridoxyliden-
glutamate, human serum albumin, and human albumin mac-
roaggregates. They concluded that the test proved to be eco-
nomical, easy, rapid, sensitive, and reliable. The test was
incorporated into the routine quality control program not only
for radiopharmaceuticals, but also for all parenteral uids and
solutions used in kit preparation within their nuclear medi-
cine department.
Rhodes and Croft (97) listed six reasons why the LAL test
is preferred over the rabbit test for pyrogen testing of radio-
pharmaceuticals and reagent kits:
1. It is more sensitive.
2. It is faster.
3. It requires smaller amounts of test material.
4. Both positive and negative controls can be performed
along with each test.
5. It does not generate radioactivity in the rabbits so it is
preferred from a radiologic safety point of view.
6. It is less expensive and easier to store.
Antineoplastic agents are another class of parenteral
medication for which the LAL test provides marked advan-
174 Chapter 2
tages over the USP rabbit test. Endotoxin is an expected con-
taminant of the enzyme L-asparaginase (34) because it is ob-
tained from cultures of E. coli ATCC 9637. However, the USP
pyrogen test cannot be used to detect endotoxin in this prepa-
ration because the rabbit is one of the species extremely sus-
ceptible to the toxic effect of the enzyme (98). L-Asparaginase
and bleomycin contain as much as 50 ng/ml endotoxin (99). It
is suspected that this contaminant is the cause of the adverse
effects seen in patients following administration of these
agents. The LAL test sensitivity characteristics aid in evaluat-
ing the techniques applied to reduce or eliminate the endo-
toxin level in these agents.
The LAL test has been used to detect the presence of bac-
terial endotoxin in 12 chemotherapeutic agents (100). Relative
concentrations of endotoxin ranging from 0.1 to 63 ng/ml were
detected in individual lots of the following drugs: L-asparagi-
nase, 5-azacytidine, bleomycin, DTIC, antinomycin D, adria-
mycin, and vinblastine. On the other hand, all lots of the fol-
lowing antineoplastic agents contained 0.1 ng/ml endotoxin:
cytosine arabinoside, cyclophosphamide, daunorubicin, vin-
cristine, and streptozotocin. The authors concluded that the
LAL test is a rapid and specic method for detection of small
amounts of bacterial endotoxin contaminating parenteral
preparations of antihumor agents.
Antibiotics are known to inhibit the LAL test at the prod-
uct concentrations used in human or animal dosages. In most
cases, however, adequate dilution of most of these products
above the minimum valid concentration (MVC) will provide
noninhibitory conditions for successful application of the LAL
test. Case et al. (101) tested 28 antibiotics with the LAL assay
to determine their noninhibitory concentrations (NICs). Most
of the antibiotics tested could be diluted to NICs above the
MVCs. Five antibiotic products presented problems. Cefaman-
dole nafate and neomycin sulfate had NICs very close to their
MVCs (1.6: 0.8 mg/ml and 0.2: 0.16 mg/ml, respectively). Poly-
myxin B and colistimethate contained too much endotoxin to
permit determination of their NICs. The NIC of tetracycline
Pyrogen Testing 175
hydrochloride was dependent on the initial concentration of
the antibiotic. If the initial concentration of tetracycline was
5 mg/ml, dilution to 0.16 mg/ml produced a noninhibitory con-
centration that was less than the MVC for tetracycline. How-
ever, a concentration of 0.5 mg/ml, when diluted, produced an
NIC that was about the same as the MVC. The reason for this
difference probably was the amount of NaOH required to ad-
just the pH of this very acidic antibiotic solution (pH 2.8). The
greater amount of base required to increase the pH of the 5.0
mg/ml product probably caused too high a sodium ion concen-
tration for the LAL test to overcome.
Other pharmaceutical preparations for which the LAL
test has proven itself as a nal product release test for pyro-
gens include LVPs (102), intravenous fat emulsions (103), iron
dextran (104), and most of the drug products listed in Table
2.11. Despite the need for dilution to eliminate the inhibitory
effects of many small-volume parenteral drug products, the
LAL test is at least equal to or more sensitive than the USP
pyrogen test.
Pharmaceuticals administered by the intrathecal route
represent a drug class most urgently in need of the LAL test
for endotoxin detection (42). Such pharmaceuticals include (a)
dyes such as methylene blue and uorescein for detecting cere-
brospinal uid (CSF) leakage, (b) contrast media for visualiza-
tion of CSF pathways, (c) cancer chemotherapeutic agents
such as methotrexate for treatment of leukemic meningitis,
(d) antibiotics such as gentamicin for septic meningitis, and
(e) radiopharmaceuticals for radionuclide cisternography, a
procedure by which a small volume of radiotracer is adminis-
tered intrathecally to study CSF dynamics by means of nu-
clear imaging devices. Endotoxin has been shown to be more
toxic following intrathecal injection compared to intravenous
injection. For example, Bennett and coworkers (105) demon-
strated in animals that instillation of endotoxin into intrathe-
cal spaces was at least 1000 times more potent in producing
a febrile response than the intravenous route.
The USP rabbit pyrogen test for intrathecal drugs has
176 Chapter 2
been shown to be insufciently sensitive to serve as a screen-
ing test for endotoxin contamination of these drugs (42). Thus,
the LAL test should replace or at least supplement the USP
pyrogen test for drugs intended to be administered into CSF.
Biologics
The FDAs Bureau of Biologics (BoB) (nowCenter for Biologics
Evaluation and Research) in 1977 published conditions under
which the LAL test can be applied as the end-product pyrogen
test for biologics (see page 124). The main requirement in-
volves validating that the LAL test and rabbit test are at least
equivalent for each product undergoing pyrogen testing. Com-
parison to the rabbit test is no longer required provided that
the test is validated to USP and FDA guidelines.
The LAL test has been used both for end-product testing
and for solving problems during the manufacturing of blood
products and plasma fractions. Expediency, sensitivity, and
quantitation of endotoxin levels are three advantages of using
the LAL test rather than the rabbit test. A comparison of the
two pyrogen tests as they are applied to various biological sub-
stances was reported by Ronneberger (65), and an example of
Ronnebergers data is given in Table 2.12.
LAL assays have been demonstrated as satisfactory for
three primary biological substances: human serum albumin
(77,106), plasma fractions (65,78,107), and vaccines (108).
However, human serumin toto inhibits the LAL gelation reac-
tion with spiked endotoxin unless modications in the test
procedure are incorporated. Human serum contains a single
protein (designated LPS-1) that inactivates LPS and inhibits
the gelation reaction (109). Other inhibitors are present in se-
rum, such as two -globulins (110) and a serumglobulin ester-
ase (111). Another variable is that the levels of substances in-
hibiting endotoxin probably vary not only from person to
person, but also in a patient during various stages of illness
associated with gram-negative infections (112).
Three methods have been reported that are capable of re-
moving these serum inhibitors. They include extracting the
Pyrogen Testing 177
T
a
b
l
e
2
.
1
2
C
o
m
p
a
r
i
s
o
n
o
f
L
A
L
a
n
d
R
a
b
b
i
t
T
e
s
t
s
o
n
V
a
r
i
o
u
s
M
a
t
e
r
i
a
l
s
L
A
L
b
d
o
s
e
R
a
b
b
i
t
a
0
.
1
m
l
N
o
.
o
f
D
o
s
e
r
a
b
b
i
t
R
a
b
b
i
t
M
a
t
e
r
i
a
l
s
a
m
p
l
e
s
m
l
/
k
g
L
V
.
L
A
L
H
e
m
a
c
c
e
l
p
l
a
s
m
a
s
u
b
s
t
i
t
u
t
e
3
5
1
0
2
3
3
2
3
3
0
P
h
y
s
i
o
l
o
g
i
c
s
a
l
i
n
e
2
1
1
0
2
1
9
2
1
9
0
A
q
u
a
d
e
s
t
.
2
5
1
0
3
2
2
3
2
2
0
A
l
b
u
m
i
n
,
h
u
m
a
n
6
5
3
2
1
4
4
3
1
3
4
1
0
G
a
m
m
a
g
l
o
b
u
l
i
n
3
1
1
8
2
3
1
6
1
5
8
F
(
a
b
)
2
-
f
r
a
g
m
e
n
t
2
7
3
8
1
9
1
2
1
5
4
P
l
a
s
m
a
p
r
o
t
e
i
n
s
o
l
u
t
i
o
n
1
8
3
4
4
1
4
9
9
5
F
i
b
r
i
n
o
g
e
n
6
1
.
5
1
5
3
3
2
F
a
c
t
o
r
X
I
I
1
0
1
2
8
2
8
0
P
r
o
t
e
i
n
a
s
e
i
n
h
i
b
i
t
o
r
9
2
1
8
1
8
9
A
n
t
a
g
o
s
a
n
I
n
t
e
r
f
e
r
o
n
6
1
4
2
5
1
1
S
t
r
e
p
t
o
k
i
n
a
s
e
3
5
1
3
3
2
5
3
0
2
N
e
u
r
a
m
i
n
i
d
a
s
e
1
2
1
4
8
6
6
2
V
a
c
c
i
n
e
s
d
i
v
.
4
5
0
.
5
1
0
2
4
2
1
2
6
2
1
2
T
o
t
a
l
3
4
5
0
.
5
1
0
1
2
7
2
5
8
1
2
3
2
2
4
4
5
P
l
a
s
m
a
h
u
m
a
n
t
r
e
a
t
e
d
c
h
l
o
-
2
8
1
1
8
1
0
5
2
3
r
o
f
o
r
m
S
o
u
r
c
e
:
R
e
f
.
6
5
.
a
R
a
b
b
i
t
2
L
A
L
P
l
a
s
m
a
1
3
.
b
L
A
L
:
P
y
r
o
g
e
n
t
.
O
t
h
e
r
s
a
m
p
l
e
s
:
4
.
178 Chapter 2
serum with chloroform (113), adjusting the plasma pH (114),
and combining the application of heat and serum dilution to
overcome the inhibiting capacity of the inhibitors (115).
Baek (116) reported an immunoelectrophoretic assay
that greatly improves the sensitivity of LAL for LPS detection
in various biological uids, including plasma. The assay
method is based on the preparation of a monospecic antibody
against coagulogen, the clottable protein formed in the LAL-
endotoxin reaction (see the section, Reaction Mechanism).
As coagulogen splits into coagulin and C-peptide, the antige-
nicity of the cleaved coagulogen is lost, and this is expressed
by a diminished migration of the protein on the rocket immu-
noelectrophoresis (RIE) plate. The method increased LPS de-
tectability in plasma by 1000 times, that is, from 1 ng LPS/
ml of plasma by visual LAL testing to 1 pg LPS/ml plasma
using RIE.
Devices
In a Federal Register notice on November 4, 1977 (42FR,
57749), the LAL test was approved by the Bureau of Medical
Devices of the FDA (now the National Center for Devices and
Radiological Health, CDRH) as a suitable test to replace the
USP pyrogen test for nal release of medical devices. As with
the biologics, the manufacturer may use the LALtest as a nal
release test for devices only after meeting four conditions:
1. Demonstrate the equivalence of the LAL test and rabbit
test for each device.
2. Document the prociency in applying the LAL test.
3. Describe LAL test methodology in detail.
4. Determine acceptance limits for the applicable device
products.
All validation data required prior to releasing devices la-
beled nonpyrogenic based on the LAL test must be kept on le
at the manufacturing site and be available for FDA inspection
(117). If a manufacturer plans to use LAL test procedures that
differ signicantly from FDA guidelines, then the manufac-
Pyrogen Testing 179
turer must submit either a 510(k) or a premarket approval
supplement.
Hundreds of device manufacturers in the United States
as well as in foreign countries have taken advantage of the
opportunity to replace the rabbit test with the LAL test. The
major barrier that must be solved for a device manufacturer
to receive regulatory approval to use the LAL test involves the
validation of the equivalency of the two pyrogen test methods.
Representatives of HIMA took the initiative to propose
guidelines (118) and conduct studies (119) to help the medical
device manufacturer successfully design and carry out proce-
dures for LAL pyrogen testing. The guidelines suggested are
summarized in Fig. 2.15. The collaborative study conducted
under HIMA auspices evaluated the pyrogenicity of E. coli
055: B5 from Difco Laboratories (Detroit, MI). Using 12 rabbit
Fig. 2.15 Schematic ow chart of the guidelines proposed by
HIMA for LAL pyrogen testing of medical devices. (From M. Weary
and F. Pearson, Pyrogen testing with Limulus amebocyte lysate,
Med. Devise Diag., Nov. 1980, Cannon Communications, Inc.).
180 Chapter 2
colonies provided by device manufacturers, contract testing
laboratories, and the FDA, the average pyrogenic dose of E.
coli 055: B5 endotoxin was found to be 0.157 ng/ml (1.57 ng/
kg for 10 ml/kg injected dose). The lower 95% condence level
was 0.1 ng/ml, meaning that if 8 rabbits were administered
this concentration of endotoxin, 4 rabbits would fail the pyro-
gen test at this dose. Therefore, any device manufacturer
wishing to use the LAL test must validate the ability of their
test procedure to detect Difco E. coli 055: B5 endotoxin at lev-
els at least equivalent to 0.1 ng/ml (1.0 ng/kg dose).
CDRH requires the following sampling guidelines to be
used for different device lot sites:
Two devices for lots less than 30
Three devices for lots of 30100
Of lots above 100, use 3%, up to a maximum of 10 devices per
lot
The recommended volume of nonpyrogenic rinsing uid per
device is 40 ml. If 10 devices are rinsed, then the total rinsing
extract pooled is 400 ml. If a rinse volume greater than 40 ml
per device is required, a more sensitive LAL end point should
be used.
Disease States
Because endotoxins are associated with gram-negative bacte-
ria, diseases caused by these bacteria conceivably can be diag-
nosed by the LAL test. A partial list of gram-negative bacteria
is given in Table 2.13 along with diseases associated with
these organisms.
Endotoxemia is a low-grade infection of the intestinal
tract caused by bacterial endotoxins. Endotoxemia can result
in endotoxic shock, which is a common cause of morbidity and
mortality in hospital patients. Detection of endotoxemia by
the LAL test was rst assessed by Fossard et al. in 1974 (120).
They concluded that the LAL test is a simple, rapid, and reli-
able method for detecting endotoxemia. Early detection per-
mits early and vigorous treatment of the infection, with the
LAL test used to monitor the effectiveness of the treatment.
Pyrogen Testing 181
Table 2.13 Gram-Negative Bacteria and Diseases Associated
with These Microorganisms
Cell shape Genus Disease
Cocci Neisseriae Gonorrhea
Meningitis
Rods Pseudomonas Wound, burn infection
Pneumonia
Eye Infection
Escherichia Gastroenteritis
Urinary tract infection
Shigella Dysentery
Proteus Urinary tract infection
Hemophilus Infantile meningitis
Chronic bronchitis
Salmonella Typhoid fever
Food poisoning
Brucella Animal infections
LALassay found high plasma endotoxin levels in patients
suffering from sepsis, malignant tumors, leukemia, and de-
compensated liver cirrhosis (121). A modication of the LAL
test was required to eliminate interference factors located in
platelet-rich plasma or serum. A simple addition of perchloric
acid to plasma in a nal concentration of 1.25% eliminated the
inhibitors. The LAL assay currently is being applied in studies
trying to determine the correlation of endotoxin levels and
various diseases.
Although not sanctioned by the FDA, the LAL test has
proven to be very useful in the diagnosis of meningitis caused
by pyrogenic radionuclide substances used in cisternography
(122) and meningitis resulting from gram-negative bacteria
(123127). The sensitivity, reliability, and rapidity of the LAL
method are vitally important because of the serious toxicity
problems associated with pyrogens and bacteria in CSF (128).
The LAL test of CSF was found to be clinically useful in neo-
nates suffering from gram-negative infection (129).
182 Chapter 2
The LAL test has been used successfully and holds impor-
tant future applications in diagnosing and monitoring such
various disease states as gingival inammation (130), bacteri-
uria (131,132), postanesthesia hepatitis (133), urinary tract
infections (134), mastitic milk (135), gram-negative sepsis re-
sulting from burns (136) and other causes (137), gonococcal
cervicitis (138), peritonitis (139), and gonorrhea (140,141).
Food
LAL testing has reached into some areas of food and drinking
water processing. LAL has been used to determine endotoxin
levels in drinking water (142,143), marine environment (144),
sugar (145), and ground beef (146). The levels of endotoxin
provide evidence of the microbial quality of the food material.
For example, LAL found 10 ng/g endotoxin for both white and
beet raw sugar, while 100 ng/g endotoxin were found in im-
ported cane raw sugar (145).
Other Applications
The LAL test has proven to be a valuable test for the detection
of endotoxin extracted from surgeons sterile latex gloves (147)
and operating nebulizers used in respiratory therapy (148).
Validation of dry heat sterilization and depyrogenation
cycles based on the destruction of endotoxin can be accom-
plished through the employment of the LAL test (149,150).
This could not be accomplished practically using the USP rab-
bit test. This has resulted in an FDA requirement for a three-
log reduction in endotoxin levels in materials being dry heat
sterilized (151).
Analogous to biological indicators used for validation and
routine monitoring of sterilization processes, there are now
endotoxin indicators that can be used in the validation and
routine control of endotoxin reduction processes (152).
MODIFICATIONS OF RABBIT AND LAL TESTS
FOR PYROGENS
Ultraltration, using a membrane having a fraction molecular
weight of 10,000, has been used to separate endotoxin contam-
Pyrogen Testing 183
ination from injectable solutions of sodium ampicillin (153).
The ltrate solution contained endotoxin, while the antibiotic
remained entrapped on the lter. Samples of the ltrate were
tested for pyrogenicity by the rabbit pyrogen test, the LAL ge-
lation test, and the chromogenic assay method. All three tests
were positive for endotoxin. Without ultraltration, the pres-
ence of sodium ampicillin interferes with rabbit, LAL, and
chromogenic detection of endotoxin (154). Ultraltration also
allows concentration of endotoxin in drug preparations, facili-
tating detecting of minute (picogram) amounts of endotoxin.
REFERENCES
1. C. M. Good and H. E. Lane, Jr., The biochemistry of pyrogens,
Bull. Parenter. Drug Assoc., 31, 116120 (1977).
2. B. D. Davis, R. Dulbecco, H. N. Eisen, and H. S. Ginsberg,
Microbiology, 3rd Edition, Harper and Row, Hagerstown,
Maryland, 1981, p. 85.
3. E. Kennedi, H. Laburn, D. Mitchell, and F. P. Ross, On the
pyrogenic action of intravenous lipid A in rabbits, J. Physiol.,
328, 361370 (1982).
4. T. Billroth, Arch. Klin. Chir., 6, 372 (1865).
5. J. Burion-Sanderson, On the process of fever, Practitioner, 16,
257 (1876).
6. G. Roussy, Gas. Hsp., Paris, 62, 372 (1889).
7. E. Centanni, Chem. Zbt., 6, 587 (1894).
8. E. Hort and W. J. Penfold, J. Hyg., 12, 361 (1912), and Proc.
Roy. Soc. Med., Ser. B,. 85, 174 (1912).
9. J. L. Jona, J. Hyg., 15, 169 (1916).
10. R. Schmidt, Muenchen Med. Wochenschr., 67, 48 (1920).
11. F. B. Siebert, Am. J. Physiol., 67, 90 (1923).
12. F. B. Siebert, The cause of many febrile reactions following
intravenous injections, Am. J. Physiol., 71, 621 (1924).
13. L. A. Rademacher, The causes of elimination of reactions after
intravenous infusions, Amer. Surg., 92, 195 (1930).
14. H. D. CoTui and M. H. Schrift, Production of pyrogen by some
bacteria, J. Lab. Clin. Med., 27, 569 (1942).
184 Chapter 2
15. F. C. Pearson, Limulus amebocyte lysate testing: comparative
methods and reagents, Sterile Pharmaceutical Manufactur-
ing: Applications for the 1990s, Vol. 2. (M. J. Groves, W. P.
Olson and M. H. Anisfeld, Eds.), Interpharm Press, 1991,
pp. 185197.
16. United States Pharmacopeia XXII/National Formulary XVII,
Mack Publishing Co., Easton, Pennsylvania, 1990, pp. 1493
1495.
17. G. R. Peroneus, Pyrogen testing of parenteral pharmaceuti-
cals, Quality Control in the Pharmaceutical Industry (M. Coo-
per, Ed.), Academic Press, New York, 1973.
18. C. J. Joubert, An automated temperature recording appara-
tus for the determination of pyrogens in intravenous solu-
tions, Lab. Anim., 14, 103106 (1980).
19. S. D. Greisman and R. B. Hornick, Comparative pyrogenic re-
activity of rabbit and man to bacterial endotoxin, Proc. Exp.
Biol., N.Y., 131, 11541158 (1969).
20. H. H. Freedman, Passive transfer of tolerance of pyrogenicity
of bacterial endotoxin, J. Exp. Med., 111, 453463 (1960).
21. D. L. Bornstein, C. Bradenberg, and W. Wood, Jr., Studies
on the pathogenesis of fever XI: Quantitative features of the
febrile response to leucocytic pyrogens, J. Exp. Med., 117,
349364 (1963).
22. P. J. Weiss, Pyrogen testing, J. Parenter. Drug Assoc., 32,
236241 (1978).
23. P. E. Mazur and M. D. McKendrick, Automated pyrogen test-
ing system, Pharm. Tech., 4, 4248 (1979).
24. British Pharmacopoeia, Volume 2, Her Majestys Stationery
Ofce, London, 1980, pp. A153154.
25. C. C. Mascoli and M. E. Weary, Limulus Amebocyte Lysate
test for detecting pyrogens in parenteral injectable products
and medical devices: Advantages to manufacturers and regu-
latory ofcials, J. Parenter. Drug Assoc., 33, 8195 (1979).
26. R. Dabbath et al., Pyrogenicity of E. coli O55: B5 endotoxin
by the USP rabbit test: A HIMA collaborative study, J. Pare-
nter Drug Assoc., 34, 212216 (1980).
27. J. G. Dare and G. A. Mogey, Rabbit responses to human
Pyrogen Testing 185
threshold doses of a bacterial pyrogen, J. Pharm. Pharmacol.,
6, 325332 (1954).
28. L. Bellentani, Cyclic and chronobiological considerations
when employing the rabbit fever test, Endotoxins and Their
Detection with the Limulus Amebocyte Lysate Test (S. W. Wat-
son, J. Levin, and T. J. Novlisky, Eds.), Alan R. Liss, New
York, 1982, pp. 329342.
29. J. Levin and F. B. Bang, The role of endotoxin in the extracel-
lular coagulation of Limulus blood, Bull. Johns Hopkins
Hosp., 115, 265274 (1964).
30. F. B. Bang, A bacterial disease of Limulus polyphemus, Bull.
Johns Hopkins Hosp., 98, 325351 (1956).
31. S. Marcus and J. R. Nelson, Tests alternative to the rabbit
bioassay for pyrogens, Dev. Biol. Stand., 34, 4555 (1977).
32. J. F. Cooper, H. D. Hochstein, and E. G. Seligman, The Limu-
lus test for endotoxin (pyrogen) in radiopharmaceuticals and
biologicals, Bull. Parenter. Drug Assoc., 26, 153162 (1972).
33. J. F. Cooper, J. Levin, and H. N. Wagner, Jr., Quantitative
comparison of in vitro and in vivo methods for the detection
of endotoxin, J. Lab. Clin. Med., 78, 138148 (1971).
34. J. F. Cooper and S. M. Pearson, Detection of endotoxin in bio-
logical products by the Limulus test, Dev. Biol. Stand., 34, 7
13 (1977).
35. U.S. Food and Drug Administration, Guideline on validation
of the Limulus Amebocyte Lysate test as an end-product endo-
toxin test for human and animal parenteral drugs, biological
products, and medical devices, Federal Register, 53, 5044
(1987), February 19, 1988.
36. Federal Register, 45, 3229932300 (1980).
37. T. Liu et al., Studies on the Limulus lysate coagulating sys-
tem, Biomedical Applications of the Horseshoe Crab (Limuli-
dae) (E. Cohen, Ed.), Alan R. Liss, New York, 1979, pp. 147
158.
38. T. Takagi et al., Amino acid sequence studies on horseshoe
crab (Tachypleus tridentatus) coagulogen and the mechanism
of gel formation, Biomedical Applications of the Horseshoe
186 Chapter 2
Crab (Limulidae) (E. Cohen, Ed.), Alan R. Liss, New York,
1979, pp. 169184.
39. M. W. Mosesson et al., Structural studies of the coagulogen
of amebocyte lysate for Limulus polyphemus, Biomedical Ap-
plications of the Horseshoe Crab (Limulidae) (E. Cohen, Ed.),
Alan R. Liss, New York, 1979, pp. 159168.
40. K. Tsuji and K. A. Steindler, Use of magnesium to increase
sensitivity of Limulus amoebocyte lysate for detection of endo-
toxin, Appl. Environ. Microbiol., 45, 13421350 (1983).
41. J. F. Cooper and M. E. Neely, Validation of the LAL test for
end product evaluation, Pharm. Tech., 4, 7279 (1980).
42. J. F. Cooper, Principles and application of the Limulus test
for pyrogen in parenteral drugs, Bull. Parenter. Drug Assoc.,
29, 122130 (1975).
43. H. D. Hochstein, The LAL test versus the rabbit pyrogen test
for endotoxin detection, Pharm. Tech., 5, 3742 (1981).
44. H. D. Hochstein, U.S. Government quality control program
for Limulus amebocyte lysate and endotoxin, Prog. Clin. Biol.
Res., 189, 221239 (1985).
45. T. J. Novitsky, J. Schmidt-Gengenbach, and J. F. Remillard,
Factors affecting recovery of endotoxin adsorbed to container
surfaces, J. Parenter. Sci. Tech., 40, 284286 (1986).
46. E. D. Guilfoyle, J. F. Yager, and S. I. Carito, The effect of re-
frigeration and mixing on detection of endotoxin in parenteral
drugs using the Limulus Amebocyte Lysate (LAL) test, J. Par-
enter. Sci. Tech., 43, 183187 (1989).
47. K. L. Melvaer and D. Fystro, Modied micromethod of the
Limulus Amoebocyte Lysate assay for endotoxin, Appl. Envi-
ron. Micro., 43, 493494 (1982).
48. J. H. Jorgensen and G. A. Alexander, Automation of the Limu-
lus amoebocyte lysate test by using the Abbott MS-2 microbi-
ology system, Appl. Environ. Microbiol., 41, 13161320
(1981).
49. T. J. Novitsky, S. S. Ryther, M. J. Case, and S. W. Watson,
Automated LAL testing of parenteral drugs in the Abbott MS-
2, J. Parenter. Sci. Tech., 36, 1116 (1982).
50. J. H. Jorgensen and G. A. Alexander, Rapid detection of sig-
Pyrogen Testing 187
nicant bacteriuria by use of an automated Limulus Amoebo-
cyte Lysate assay, J. Clin. Microbiol., 16, 587589 (1982).
51. B. Diller et al., Detection of endotoxin in blood and other speci-
mens by evaluation of photometrically registered LAL-reac-
tion kinetics in microliter plates, Endotoxins and Their Detec-
tion with the Limulus Amebocyte Lysate Test (S. W. Watson,
J. Levin, and T. J. Novitsky, Eds.), Alan R. Liss, New York,
1982, pp. 385392.
52. M. Harada, T. Morita, and S. Iwanaza, A new assay method
for bacterial endotoxins using horseshoe crab clotting enzyme,
J. Med. Enzymol., 3, 43 (1978).
53. K. Tsui, P. A. Martin, and D. M. Bussey, Automation of chro-
mogenic substrate Limulus Amebocyte Lysate assay method
for endotoxin by robotic system, Appl. Environ. Microbiol., 48,
550555 (1984).
54. J. F. Cooper, Chromogenic substrates offer realistic alterna-
tive to LAL gel endpoint, Partic. Micro. Contr., 2, 3233
(1983).
55. G. K. Lindsay, P. F. Roslansky, and T. J. Novitsky, Single-
step, chromogenic Limulus Amebocyte Lysate assay for endo-
toxin, J. Clin. Microbiol., 27, 947951 (1989).
56. T. J. Novitsky, LAL methodology: The choice is yours, Med.
Dev. Diagn. Ind., 4953 (Jan. 1984); updated periodically in
LAL Update, Associates of Cape Cod, Inc., Woods Hole, Mas-
sachusetts.
57. M. E. Weary, Understanding and setting endotoxin limits, J.
Parenter. Sci. Tech., 44, 1618 (1990).
58. U.S. Food and Drug Administration, Licensing of Limulus
amebocyte lysate: Use as an alternative for rabbit pyrogen
test, Federal Register, 42, 57749 (1977).
59. Parenteral Drug Association, Parenteral Drug Association Re-
sponse to FDA Draft Guideline for the Use of Limulus
Amebocyte Lysate, Information Bulletin 3, March 1980.
60. U.S. Food and Drug Administration, Guideline for Validation
of Limulus Amebocyte Lysate Test as an End Product Endo-
toxin Test for Human and Animal Parenteral Drugs, Biologi-
cal Products and Medical Devices, December 1987.
188 Chapter 2
61. K. Tsuji, K. A. Steindler, and S. J. Harrison, Limulus
Amebocyte Lysate assay for endotoxin for detection and quan-
titation of endotoxin in a small-volume parenteral product,
Appl. Environ. Microbiol., 40, 533538 (1980).
62. A. J. Schmitz, Development of endotoxin limits for USP arti-
cles, articles in-process and potential articles, Stimuli to the
Revision Process, Pharm. Forum, 29472971 (1987).
63. R. J. Elin and S. M. Wolff, Nonspecicity of the Limulus
Amebocyte Lysate test: Positive reaction with polynucleotides
and proteins, J. Infect. Dis., 128, 3 (1973).
64. J. D. Sullivan and S. W. Watson, Factors affecting the sensi-
tivity of Limulus lysate, Appl. Microbiol., 28, 10231026
(1974).
65. H. J. Ronneberger, Comparison of the pyrogen tests in rabbits
and with Limulus lysate, Dev. Biol. Stand., 34, 2732 (1977).
66. G. Nyerges and S. Jaszovszky, Reliability of the rabbit pyro-
gen test and of the Limulus test in predicting the pyrogenicity
of vaccines in man, Acta Micro. Acad. Sci. Hung., 28, 235
243 (1981).
67. R. E. Wachtel and K. Tsuji, Comparison of Limulus amebocyte
lysates and correlation with the United States Pharmaco-
poeia pyrogen test, Appl. Environ. Microbiol., 33, 12651269
(1977).
68. C.W. Twohy, M. L. Nierman, A. P. Duran, and T. E. Munson,
Comparison of Limulus amebocyte lysates from different
manufacturers, J. Parenter. Sci. Tech., 37, 9396 (1983).
69. A. Widfeuer et al., Use of Limulus assay to compare the bio-
logical activity peptidoglycan and endotoxin, A. Immutaetsf-
orsch. 149, 258264 (1975).
70. K. Brunson and D. W. Watson, Limulus amebocyte lysate re-
action with streptococcal pyrogenic endotoxin, Infect. Im-
mun., 14, 12561258 (1970).
71. M. Suzuki et al., Gelation of Limulus lysate by synthetic dex-
tran derivatives, Microbiol. Immunol., 21, 419425 (1977).
72. D. H. Fine et al., Limulus lysate activity of lipoteichoic acids,
J. Dent. Res., 56, 15001501 (1977).
73. M. Platica, W. Harding, and V. P. Hollander, Dithiols stimu-
Pyrogen Testing 189
late endotoxin in the Limulus reaction, Experientia, 34, 1154
1155 (1978).
74. K. Ikemura, K. Ikegami, T. Shimazu, T. Yoshioka, and T. Sug-
imoto, False-positive results in Limulus test caused by Limu-
lus amebocyte lysatereactive material in immunoglobulin
products, J. Clin. Microbiol., 27, 19651968 (1989).
75. N. Ohno, Y. Emori, T. Yadomae, K. Saito, A. Masuda, adn S.
Oikawa, Reactivity of Limulus amoebocyte lysate towards
(1 3)-a -D-glucans, Carbohydr. Res., 207, 311318 (1990).
76. J Kambayashi, M. Yokota, M. Sakon, E. Shiba, T. Kawasake,
T. Mori, M. Tsuchiya, M. Oishi, and S. Matsuura, A novel en-
dotoxin-specic assay by turbidimetry with Limulus amoebo-
cyte lysate containing beta-glucan, J. Biochem. Biophys.
Meth., 22, 93100 (1991).
77. F. C. Pearson and M. Weary, The signicance of Limulus
Amebocyte Lysate test specicity on the pyrogen evaluation
of parenteral drugs, J. Parenter. Drug Assoc., 34, 103108
(1980).
78. A. Gardi and G. Arpagaus, The Limulus Amebocyte Lysate
test: Auseful tool for the control of plasma fractions, Dev. Biol.
Stand., 34, 2126 (1977).
79. D. Fumarola and E. Jirillo, Limulus test, parenteral drugs
and biological products: An approach, Dev. Biol. Stand., 34,
97100 (1977).
80. C. C. Mascoli and M. E. Weary, Applications and advantages
of the Limulus Amebocyte Lysate (LAL) pyrogen test for par-
enteral injectable products, Prog. Clin. Biol. Res., 29, 387
402 (1979).
81. M. Weary and B. Baker, Utilization of the Limulus Amebocyte
Lysate test for pyrogen testing large-volume parenterals, ad-
ministration sets, and medical devices, Bull. Parenter. Drug
Assoc., 31, 127133 (1977).
82. C. C. Mascoli, Limulus Amebocyte Lysate (LAL) test for de-
tecting pyrogens in parenteral injectables and medical de-
vices, J. Pharm. Belg., 34, 145155 (1979).
83. Current practices in endotoxin and pyrogen testing in biotech-
nology, The Quality Assurance/Quality Control Task Group,
190 Chapter 2
Parenteral Drug Association, J. Parenter. Sci. Tech., 44, 39
45 (1990).
84. J. vanNoordwijk and Y. DeJong, Comparison of the LAL test
with the rabbit test: False positives and false negatives, Dev.
Biol. Stand., 34, 3943 (1977).
85. K. Z. McCullough and S. A. Scolnick, Effect of semisynthetic
penicillins on the Limulus Lysate test, Antimicrob. Ag. Che-
mother., 9, 856 (1976).
86. K. L. Smith, Sequential analysis applied to biological control
tests for pharmacopoeia substances, Quantitative Methods in
Pharmacology (J. deJonge, Ed.), North Holland, Amsterdam,
1961, pp. 4955.
87. J. VanNoordwijk and Y. DeJong, Comparison of the Limulus
test for endotoxin with the rabbit test for pyrogens of the Eu-
ropean pharmacopoeia, J. Biol. Stand., 4, 131139 (1976).
88. J. D. Sullivan and S. W. Watson, Inhibitory effect of heparin
on the Limulus test for endotoxin, J. Clin. Microbiol., 2, 151
(1975).
89. J. D. Sullivan and S. W. Watson, Purication and properties
of the clotting enzyme from Limulus lysate, Biochem. Bios-
phys. Res. Commun., 66, 848855 (1975).
90. J. F. Cooper, Resolving LAL test interferences, J. Parenter.
Sci. Tech., 44, 1315 (1990).
91. K. Z. McCullough, Variability in the LAL test, J. Parenter.
Sci. Tech., 44, 1921 (1990).
92. J. F. Cooper, Ideal properties of a LALreagent for pharmaceu-
tical testing, Bacterial Endotoxins: Structure, Biomedical Sig-
nicance, and Detection with the Limulus Amebocyte Lysate
Test (S. W. Watson, J. Levin, and T. J. Novitsky, Eds.), Alan
R. Liss, New York, 1985, pp. 241249.
93. C. W. Twohy, A. P. Duran, and T. E. Munson, Endotoxin con-
tamination of parenteral drugs and radiopharmaceuticals as
determined by the Limulus Amebocyte Lysate method, J. Par-
enter. Sci. Tech., 30, 190201 (1984).
94. T. J. Novitsky, J. Schmidt-Gengenbach, and J. F. Remillard,
Factors affecting recovery of endotoxin adsorbed to container
surfaces, J. Parenter. Sci. Tech., 40, 284286 (1986).
Pyrogen Testing 191
95. A. K. Highsmith, R. L. Anderson, and J. R. Allen, Application
of the Limulus Amebocyte Lysate assay in outbreaks of pyro-
genic reaction associated with parenteral uids and medical
devices, Endotoxins and Their Detection with the Limulus
Amebocyte Lysate Test (S. W. Watson, J. Levin, and T. J. No-
vitsky, Eds.), Alan R. Liss, New York, 1982, pp. 287299.
96. C. A. DeMurphy and T. R. Aneiros,
99m
Tc-radiopharmaceuti-
cals and the Limulus test, J. Parenter. Drug Assoc., 34, 268
271 (1980).
97. B. A. Rhodes and B. Y. Croft, Basics of Radiopharmacy, C. B.
Mosby, St. Louis, Missouri, 1978, p. 81.
98. H. F. Oettgen et al., Toxicity of E. coli L-asparaginase in man,
Cancer, 35, 253278 (1970).
99. D. Fumarola, Possible endotoxin contamination in some anti-
tumor agent preparations: Study with Limulus Amebocyte
Lysate test II, Farmaco, 32, 444448 (1977).
100. S. E. Siegel, R. Nachum, S. Leimbrock, and M. Karon, Detec-
tion of bacterial endotoxin in anti-tumor agents, Cancer Treat.
Rep., 60, 915 (1976).
101. M. J. Case, S. S. Ryther, and T. J. Novitsky, Detection of endo-
toxin in antibiotic solutions with Limulus amoebocyte lysate,
Antimicrobiol Agents Chemother., 23, 649652 (1983).
102. R. Nandan and D.R. Brown, An improved in vitro pyrogen test
to detect picograms of endotoxin contamination in intrave-
nous uids using Limulus amebocyte lysate, J. Lab. Clin.
Med., 89, 910918 (1977).
103. R. J. Guzman and H. Kuo, Pyrogen testing of an intravenous
fat emulsion, Prog. Clin. Biol. Res., 29, 333337 (1979).
104. S. L. Gafn, Endotoxin determination in viscous opaque solu-
tions of iron dextran by Limulus amebocyte lysate, Prog. Clin.
Biol. Res., 29, 221227 (1979).
105. I. L. Bennett, Jr., R. G. Petersdorf, and W. R. Keene, Patho-
genesis of fever: Evidence for direct cerebral action of bacte-
rial endotoxins, Trans. Assoc. Am. Phy., 70, 64 (1957).
106. F. C. Pearson et al., Applications of the Limulus Amebocyte
Lysate assay to testing of HSA: Symposium on sterilization
microbiology.
192 Chapter 2
107. H. Ronneberger, LAL assay for plasmaproteins, Endotoxins
and Their Detecting with the Limulus Amebocyte Lysate Tests
(S. W. Watson, J. Levin, and T. J. Novitsky, Eds.), Alan R.
Liss, New York, 1982, pp. 261267.
108. S. C. Rastogi, H. D. Hochstein, and E. G. Seligimann, Statisti-
cal determination of endotoxin content in inuence vaccine by
the LAL test, J. Clin. Microbiol., 2, 144 (1977).
109. K. J. Johnson, P. A. Ward, S. Goralnick, and M. J. Osborn,
Isolation from human serum of an inactivator of bacterial li-
popolysaccharide, Am. J. Pathol., 88, 559 (1977).
110. R. C. Skarnes, The inactivation of endotoxin after interaction
with certain proteins of normal serum, Ann. N.Y. Acad. Sci.,
133, 644622 (1966).
111. F. S. Rosen, R. C. Skarnes, and M. Landy, Inactivation of en-
dotoxin by humoral component III. Role of divalent cation and
dialyzable component, J. Exp. Med., 108, 701 (1958).
112. J. Levin, The Limulus test and bacterial endotoxins: Some
perspectives, Endotoxins and Their Detection with the Limu-
lus Amoebocyte Lysate Test (S. W. Watson, J. Levin, and T. J.
Novitsky, Eds.), Alan R. Liss, New York, 1982, pp. 724.
113. J. Levin, P. A. Tomasulo, and R. S. Oser, Detection of endo-
toxin in human blood and demonstration of an inhibitor, J.
Lab. Clin. Med., 75, 903 (1970).
114. R. B. Reinhold and J. Fine, A technique of quantitative mea-
surement of endotoxin in human plasma, Proc. Soc. Exp. Biol.
Med., 137, 334 (1971).
115. M. S. Cooperstock, R. P. Tucker, and J. V. Baublis, Possible
pathogenic role of endotoxin in Reyes syndrome, Lancet, 1,
1272 (1975).
116. L. Baek, New, sensitive rocket immunoelectrophoretic assay
for measurement of the reaction between endotoxin and Lim-
ulus amoebocyte lysate, J. Clin. Microbiol., 17, 10131020
(1983).
117. V. C. Ross, LAL testing of medical devices: A regulatory up-
date, Med. Dev. Diagn. Ind., 3739, 91 (March, 1984).
118. Guidelines for the Use of Limulus Lysate Test (LAL) for Pyro-
Pyrogen Testing 193
gen Testing of Medical Devices, HIMAReport No. 78-8, Health
Industry Manufacturers Association, Washington, D.C., 1978.
119. HIMA Collaborative Study for the Pyrogenicity Evaluation of
a Reference Endotoxin by the USP Rabbit Test, HIMA Docu-
ment Series, 1, No. 7, Health Industry Manufacturers Associ-
ation, Washington, D.C., 1979.
120. D.P. Fossard, V. V. Kakkar, and P. A. Elsey, Assessment of
Limulus test for detecting endotoxaemia, Br. Med. J., 2, 465
(1974).
121. T. Obayashi et al., New Limulus Amoebocyte Lysate test for
endotoxaemia, Lancet, 1, 289 (1982).
122. J. F. Cooper and J. C. Harbert, Endotoxin as a cause of aseptic
meningitis after radionuclide cisternography, J. Nucl. Med.,
16, 809813 (1975).
123. J. H. Jorgensen and J. C. Lee, Rapid diagnosis of gram-nega-
tive bacterial meningitis by the Limulus endotoxin assay, J.
Clin. Microbiol., 7, 12 (1978).
124. N. Clumeck, Limulus test and meningitis, Br. Med. J., 1, 777
(1977).
125. D. Dyson and G. Cassady, Use of Limulus lysate for detecting
gram-negative neonatal meningitis, Pediatrics, 58, 105
(1976).
126. S. Ross et al., Limulus lysate test for gram-negative bacterial
meningitis, JAMA, 233, 1366 (1975).
127. T. L. Swelle, L. M. Dunkle, and L. Blair, Correlation of cere-
brospinal uid endotoxin-like activity with clinical and labo-
ratory variables in gram-negative bacterial meningitis in chil-
dren, J. Clin. Microbiol., 25, 856858 (1987).
128. G. H. McCracken and L. D. Sarff, Endotoxin in cerebrospinal
uid, JAMA, 235, 617 (1976).
129. M. C. Kelsey, A. P. Lipscomb, and J. M. Mowles, Limulus
Amoebocyte Lysate endotoxin test: An aid to the diagnosis in
the septic neonate?. J. Infect., 4, 6973 (1982).
130. L. Shapiro et al., Endotoxin determinations in gingival in-
ammation, J. Periodontal Res., 43, 59 (1972).
131. S. H. Jorgensen et al., Rapid detection of gram-negative bacte-
194 Chapter 2
riuria by use of the Limulus endotoxin assay, J. Appl. Micro-
biol., 26, 3842 (1973).
132. R. Nachumand R. N. Berzofsky, Chromogenic Limulus Amoe-
bocyte Lysate assay for rapid detection of gram-negative bac-
teriuria, J. Clin. Microbiol, 21, 759763 (1985).
133. C. Lamanto et al., Horseshoe crabs, endotoxemia and post-
anesthesia hepatitis, N.Y. State J. Med., 2145 (Nov. 1974).
134. J. M. Jorgensen and P.M. Jones, Comparative evaluation of
the Limulus assay and the direct gram stain for the detection
of signicant bacteriuria, Am. J. Clin. Pathol., 63, 142 (1975).
135. G. Hartmann and A. Saran, Application of the Limulus
Amebocyte Lysate test to the detection of gram-negative bac-
terial endotoxin in normal and mastitic milk, Res. Vet. Sci.,
20, 342 (1976).
136. N. Harris and R. Feinstein, A new Limulus assay for the de-
tection of endotoxin, J. Trauma, 17, 714718 (1977).
137. F. C. Pearson, J. Dubczak, M. Weary, G. Bruszer, and G. Do-
nohue, Detection of endotoxin in the plasma of patients with
gram-negative bacterial sepsis by the Limulus Amoebocyte
Lysate Assay, J. Clin. Microbiol., 21, 865868 (1985).
138. V. A. Spagna, R. B. Prior, and G. A. Sawaya, Sensitivity, speci-
city, and predictive values of the Limulus lysate assay for
detection of exclusion of gonococcal cervicitis, J. Clin. Micro-
biol., 16, 7781 (1982).
139. M. D. Clayman, A. Raymond, D. Cohen, C. Moftt, C. Wolff,
and E. G. Neilson, The Limulus Amebocyte Lysate assay: A
rapid and sensitive method for diagnosing early gram-nega-
tive peritonitis in patients undergoing continuous ambulatory
peritoneal analysis, Arch. Intern. Med., 147, 337340 (1987).
140. B. L. Hainer, P. Danylchuk, J. Cooper, and C. W. Weart, Lim-
ulus lysate assay in detection of gonorrhea in women from a
low-incidence population, Am. J. Obstet. Gynecol., 144, 6771
(1982).
141. V. A. Spagna and R. B. Prior, Detection of gonorrhea by Limu-
lus lysate assay, Am. J. Obstet. Gynecol., 146, 474475 (1983).
142. J. H. Jorgensen, J. C. Lee, and H. R. Pahren, Rapid detection
Pyrogen Testing 195
of bacterial endotoxin in drinking water and renovated waste
water, J. Appl. Microbiol., 32, 347351 (1976).
143. T. M. Evans, J. E. Schillinger, and D. G. Stuart, Rapid deter-
mination of bacteriological water quality by using Limulus ly-
sate, Appl. Environ. Microbiol., 35, 376382 (1978).
144. S. W. Watson et al., Determination of bacterial number and
biomass in the marine environment, Appl. Environ. Micro-
biol., 33, 940946 (1977).
145. G. Haskae and R. Nystrand, Determination of endotoxins in
sugar with the LAL test, Appl. Environ. Microbiol., 38, 1078
1080 (1979).
146. J. M. Jay, The Limulus lysate endotoxin assay as a test of
microbial quality of ground beef, J. Appl. Bacteriol., 43, 99
100 (1977).
147. R. Kure, H. Grendahl, and J. Paulssen, Pyrogens from sur-
geons sterile latex gloves, Acta Path. Microbiol. Immunol.
Scand. Sect. B, 90, 8588 (1982).
148. D. J. Reinhardt et al., LAL and direct sampling methods for
surveillance of operating nebulizers, Appl. Environ. Micro-
biol., 42, 850855 (1981).
149. M. J. Akers, K. M. Ketron, and B. R. Thompson, F value re-
quirements for the destruction of endotoxin in the validation
of dry heat sterilization/depyrogenation cycles, J. Parenter.
Sci. Tech., 36, 2327 (1982).
150. K. Tsuji and S. J. Harrison, Dry heat destruction of lipopoly-
saccharide: Dry heat destruction kinetics, Appl. Environ. Mi-
crobiol., 36, 710 (1978).
151. U.S. Food and Drug Administration, Guideline on Sterile
Drug Products Produced by Aseptic Processing, 1987.
152. LAL Users Group, Preparation and use of endotoxin indica-
tors for depyrogenation process studies, J. Parenter. Sci.
Tech., 43, 109112 (1989).
153. S. Takahashi, S. Yano, Y. Nagaoka, K. Kawamura, and S. Mi-
nami, A highly sensitive pyrogen test for antibiotics I: Detec-
tion of trace amounts of endotoxin in injectable sodium ampi-
cillin preparations, J. Pharm. Sci., 72, 739742 (1983).
154. P. M. Newsome, Penicillins and the Limulus Amebocyte Ly-
196 Chapter 2
sate test for endotoxin, J. Pharm. Pharmacol. 29, 204206
(1977).
155. K. Zink-McCullough and C. Weidner-Loeven, Variability in
the LAL Test: Comparison of Three Kinetic Methods for the
Testing of Pharmaceutical Products, Journal of Parenter. Sci.
Tech., 6972 (MayJune 1992).
3
Particulate Matter Testing
INTRODUCTION
No quality control test, parenteral or nonparenteral, presents
more difculties for quality control specialists than inspection
and analysis of injectable solutions for the presence of foreign
particulate matter. The oldest, yet most commonly used, test
for particulate matter evaluation involves human visual ex-
amination. Such examination is subjective, time consuming,
and limited in the types of parenteral products and containers
that can be inspected. This has stimulated many studies re-
garding ways of not only improving efciency of human in-
spection, but also developing and improving methods of de-
tecting particulate matter electronically.
The U.S. Pharmacopeia (USP) requirement for injectable
products species that each nal container of injection be
subjected individually to a physical inspection, whenever the
nature of the container permits, and that every container
whose contents show evidence of contamination with visible
foreign material be rejected (1). Additional specications are
required for subvisible particulate matter content and analy-
197
198 Chapter 3
sis in large-volume injections (LVIs) for single-dose infusion
and in small-volume parenterals (2).
The European Pharmacopeia (EP) contains sections on
subvisible particulate contamination and visible particulate
contamination (3).
Why are injectable solution products to be free of visible
evidence of particulate matter? Primarily, lack of particulate
matter conveys a clean, quality product, indicative of the high-
quality standards employed by the product manufacturer.
Moreover, in recent years, particulate matter has become
known as a potential hazard to the safety of the patient under-
going parenteral therapy. While there still seems to be a lack
of sufcient clinical data to incriminate particles as producers
of signicant clinical complications during parenteral ther-
apy, it is a universal belief in the health care eld that particu-
late matter does present a clinical hazard and must be absent
from the injectable solution.
The aim of this chapter is to concentrate on particle test-
ing methods in the quality control analysis of parenteral solu-
tions. Two primary methods of particulate analysisvisual
inspection and electronic particle countingare discussed in
detail. Barbers recent edition (4) represents an encyclopedic
reference for particulate matter testing.
BACKGROUND OF PARTICULATE MATTER CONCERNS
IN PARENTERAL PRODUCTS
It is interesting to realize that all the attention given today to
the problems and analysis of particulate matter in parenteral
products did not exist before the 1940s. After the inclusion of
the rst injectable product in the USP (12th edition) in 1942,
Godding (5) was the rst individual to publish an article con-
cerning the need for standards in the visual inspection of par-
ticulate matter. The 13th edition of the USP(6) gave a detailed
method for inspecting an injectable solution against a white-
and-black background using a light intensity between 100 and
350 footcandles at a distance of 10 inches. Interestingly, the
Particulate Matter Testing 199
method described in the 13th edition is still widely used in
manual inspections for evidence of visible particulate matter.
The rule-of-thumb standard that a person with 20/20
vision under inspection conditions should be able to detect ap-
proximately 30-m particles came from a report by Brewer
and Dunning (7). This detection limit has persevered since
1957, although later research suggested that inspectors
should actually be able to see particles in the size range of 20
m (8).
In the early 1950s, a number of reports began citing evi-
dence of biological hazards produced by foreign injected mate-
rials. Among the materials found to cause pulmonary granulo-
mata or emboli were cotton bers (9) and cellulose (10). Glass
particles and their potential hazard was studied by Brewer
and Dunning (7) and later by Gnadinger (11), but no evidence
of foreign body reactions in animals were found. These and
other reports led to the classic works done by Garvan and Gun-
ner published in 1963 and 1964 (12,13). These Australian phy-
sicians showed that foreign body granulomas could be pro-
duced experimentally in the lungs of rabbits following the
administration of 500 ml saline solution contaminated with
visible particulate matter. Most commercial intravenous solu-
tions inspected contained particle contamination and the
source of most of the particles was attributed to the rubber
closure. For every 500 ml of particle-contaminated intrave-
nous solution injected into a rabbit, 5000 granulomas ap-
peared in the lungs. Garvan and Gunner further found that
similar granulomas appeared in the postmortem examina-
tions of the lungs of patients receiving large volumes of intra-
venous uids. Their comments included the possibility that
postoperative pulmonary infarction was a result of particulate
thrombosis. The repercussions of Garvan and Gunners re-
ports have stimulated numerous studies that continue to this
day on the analysis and potential clinical hazards of particu-
late matter.
A collaborative study conducted by the Pharmaceutical
Manufacturers Association (14) involved the intravenous in-
200 Chapter 3
jection of varying quantities and sizes of inert polystyrene
spheres into hundreds of rats, then performing necropsies at
various periods of time from 1 hour to 28 days following injec-
tion. The results were as follows:
1. Of 18 rats injected with 8 10
6
particles per kilogram at
a particle size of 40 m, 13 died within 5 minutes.
2. Rats showed normal blood studies, organ weights, and
pathologic criteria after being injected with either 8 10
6
particles at a size of 0.4 to 10 m or 4 10
5
particles per
kilogram of 40 m particle.
3. Particles in the 4-m size range were found in the lung,
liver, and spleen.
4. Particles in the 10-m size range were found in the lung
primarily, although particles were found in ve other or-
gans.
5. Particles in the 40-m size range were found in the lungs
and myocardial tissue.
It was concluded that nonreactive particles administered
intravenously over a broad size range and up to dosages that
produced death were without clinical or tissue toxicity. Much
disagreement resulted over this conclusion, especially because
of the articial nature of the type of particle studied. However,
the same size-dependent localization of particles in different
organs was found in the case of glass particles derived from
breaking the necks of glass ampules (15). Large particles (20
m) were retained mostly in the lungs of mice, while smaller
particles (510 m) were found in the liver, spleen, and kid-
ney. No glass particles were found in the brain.
The potential hazard of particulate matter has been im-
plied in a number of reports, two of which are cited here.
In a study of 173 patients undergoing cardiac catherization
and/or surgery, 14 (8%) had ber emboli in routine autopsy
sections (16). The embolized ber often resulted in narrowing
or occlusion of the involved blood vessel. Three cases of myo-
cardial infarction were associated with embolic bers. Fibers
were believed to have originated from various materials used
in surgery and from drug solutions. It was concluded that par-
ticulate matter is a hazard, and all steps must be taken to
prevent its inadvertent administration.
A second critical report implicating the hazards of partic-
ulate matter was the work published by DeLuca et al. (17). In
a repeated double-blind study of 146 patients, a signi-
cant reduction in the incidence of infusion phlebitis was seen
when patients were administered intravenous uids ltered
through an in-line 0.45-m lter. Other studies, as reviewed
by Turco and King (18), have supported this nding.
Only ancillary reports have appeared in the literature in
the past few years regarding new ndings on the clinical haz-
ards of particulate matter. These include a report by Stein and
Vu (19) that a piece of rubber, apparently cored out of the rub-
ber closure of a vial, caused some interference during cataract
surgery.
Barber (20) cites the review by Pesko (21) as the most
recent and best focused review of the literature on the hazards
of particulate matter. Some interesting facts from this article
include the following:
Size, number, rate of introduction, and type of particle enter-
ing the bloodstream will all contribute to what harm, if
any, the particle(s) actually produce. Some particles
might cause allergic reactions.
The health condition of the person receiving solutions con-
taining particulate matter also greatly matters with re-
spect to potential harm of these particles.
Some particles will cause an inammatory response. The po-
tential harm of this inammatory response depends on
where the particles end up in the body and, if they are
located in a vital organ, what is that organs capacity to
compensate for the insult caused by the foreign matter.
Freedom from visible evidence of particulate matter is a
basic, essential characteristic of injectable products. Such a
characteristic imparts three signicant qualities to the
product:
202 Chapter 3
1. Signicance to the manufacturerlack of particulate
matter indicates good production technique and a high-
quality product.
2. Signicance to the userlack of particulate matter indi-
cates a clean product that is safe to the patient and con-
veys high-quality standards employed by the manufac-
turer of the product.
3. Clinical signicancelack of particulate matter indicates
the lack of potential hazards resulting from particles en-
tering the circulatory system, although controversy still
exists regarding the hazards of particulate matter (23).
NATURE AND SOURCE OF PARTICULATES
Anything that directly or indirectly comes in contact with a
parenteral solution, including the solvent and solutes compos-
ing the solution itself, represents a potential source of particu-
late contamination. Table 3.1 lists common sources of particu-
lates found in parenteral solutions.
The type and approximate size range of particulates
found in parenteral products are listed in Table 3.2. The small-
est capillary blood vessels are considered to have a diameter
of approximately 7 m. Thus, all particles having a size equal
to or greater than 7 m can conceivably become entrapped in
and occlude a blood capillary. Most particulates, as seen in
Table 3.2, potentially can be this size and, obviously, represent
a hazard to the health of a patient administered parenteral
medications containing these contaminants.
It seems that, regardless of whatever painstaking proce-
dures are undertaken to eliminate particle contamination,
parenteral solutions always contain a certain degree of partic-
ulate matter. It is always an uncertainty whether the particles
originated during the manufacturing and packaging process
or were introduced during the analysis of the solution for the
presence of particulates. The emphasis on technique in the
analysis of particulate matter has been stressed by Draftz and
Graf (25), McCrone (26), and Barber (27). It is imperative that
Table 3.1 Common Sources
of Particulate Matter
1. Chemicals
a. Undissolved substances
b. Trace contaminants
2. Solvent impurities
3. Packaging components
a. Glass
b. Plastic
c. Rubber
d. Intravenous
4. Environmental contaminants
a. Air
b. Surfaces
c. Insect parts
5. Processing equipment
a. Glass
b. Stainless steel
c. Rubber
d. Rust
6. Filter bers
7. People
a. Skin
b. Hair
c. Gowning
particles seen in solutions have not originated during the par-
ticle measurement and identication procedures.
THE REALITY OF PARTICULATE MATTER
CONTAMINATION IN PARENTERAL SOLUTIONS
While it is desirable to prepare and use parenteral products
completely free from particulate matter, it must be admitted
that this ideal state is not possible. All parenteral products
contain some level of particulate matter contamination. The
question is, How many particles, of what type, and what
size?
204 Chapter 3
Table 3.2 Type and Approximate Size Range
of Extraneous Materials Reported in Parenteral
Solutions
Approximate
Material size range (m)
Glass 1
Metal 1
Rubber 1 to 500
Starch 1
Zinc oxide 1
Whiting 1
Carbon black 1
Clay 1
Diatoms 1 to 5
Bacteria 2
Fungi and fungal spores 20
Insect parts 20
Cellulose bers 1 to 100
Trichomes 10
Miscellaneous crystalline material 1
Talc 1
Asbestos bers 1 to 100
Unidentied bers 1
Source: Ref. 24.
Barber (4) addresses this question throughout his book.
Below are excerpts taken from his book.
Particulate matter present in parenteral solutions and
medical devices has been an issue in the pharmaceutical
industry since the introduction of injectable preparations
and remains unavoidable, even with todays well con-
trolled manufacturing processes. (p. 2)
Correctly or incorrectly, the particulate matter bur-
den of a product has been taken by some healthcare prac-
titioners, academic investigators, and regulatory person-
nel as an indicator of overall product quality. This is
unfortunate, since particulate matter is, realistically,
only a single parameter by which product suitability or
conformity may be judged. (p. 2)
The practice content of IV uids in plastic containers
has been repeatedly found to be lower than that of glass
containers. (p. 14)
There are three very important concepts embodied
in the denitions provided by the USP, JP, and BP:
1. Particulate matter currently exists at extremely low
levels in injectable products so that there is no de-
monstrable evidence of adverse patient effects.
2. The material cannot be monotypic, but rather results
from a variety of sources inherent in a GMP-con-
trolled production process.
3. The material is not amendable to chemical analysis
due to the small mass that it represents and its het-
erogeneous composition. Thus, the appropriate ana-
lytical enumeration of this material must be sensitive
physical tests that detect size and quantitate the ma-
terial based on its optical properties. (pp. 2021)
The occurence of low numbers of heterogeneously
sourced particles is inevitable in the manufacture of in-
jectable products and medical devices. (p. 22)
The USP requires that injectables be essentially
free of visible particulate matter. The allowable particle
burden of units tested must ultimately be judged with re-
spect to the acceptable small quantity of visible particu-
late matter that may be present in units produced by par-
enteral manufacture under cGMP conditions. The most
important aspect of the visual inspection procedure is the
detection of any particulate matter that is related to solu-
tion degradation or any particulate matter present in suf-
cient quantity or of sufcient size to constitute a non-
GMP condition. (pp. 261262)
Despite rigorous cleaning procedures either at the
vendors plant or in-house, glass vials and stoppers occa-
sionally will bear or contain a single visible particle fol-
lowing lling, stoppering or lyophilization, and these par-
206 Chapter 3
ticles may become free in the solution. Such single visible
particles of random isolated occurrence are analogous to
the allowable particle burden under USP 788. A
unit bearing such an isolated single visible particle must
be rejected. On the basis of the low level of occurrence
of such visible particulates in the batch of manufactured
materials from which the unit came, however, the batch
may be still be considered essentially, substantially, and
practically free of visible particulate material. (p. 262)
It is important for the pharmaceutical industry and
regulators to recognize that there are no particle-free par-
enteral solutions. There is, further, no evidence of any
patient issue related to infusion of a small number of
inert particles with the current or previous USP limits.
(p. 273)
PARTICULATE MATTER STANDARDS
The rst reference to particulate matter in the USP occurred
in the eighth edition in 1905 (28,29). Diphtheria Antitoxin, a
hypodermic injection product, was described as a transparent
or slightlyturbidliquid. Not until 1936, inthe National Formu-
lary(NF), sixthedition, wasthetermclearness denedfor par-
enteral products (30): Aqueous Ampul Solutions are to be clear;
that is, when observed over a bright light, they shall be sub-
stantially free from precipitate, cloudiness or turbidity, specks
of bers, or cotton hairs, or any undissolved material (24).
The words substantially free caused interpretive difcul-
ties; thus, in 1942, the NF, seventh edition, provided a deni-
tion: Substantially free shall be construed to mean a prepara-
tion which is free from foreign bodies that would be readily
discernible by the unaided eye when viewed through a light
reected from a 100-watt mazda lamp using as a medium a
ground glass and a background of black and white. It was
also in 1942 that the 12th edition of the USP contained its
rst particulate matter standard:
Appearance of Solution or Suspension Injections which
are solutions of soluble medicaments must be clear, and
free (note the absence substantially) of any turbidity or
undissolved material which can be detected readily with-
out magnication when the solution is examined against
black and white backgrounds with a bright light reected
from a 100-watt mazda lamp or its equivalent.
The requirement that every injectable product in its nal
container be subjected individually to visual inspection ap-
peared in the 13th edition of the USP (5). This requirement
has remained essentially unchanged; as the 25th edition
states: Good pharmaceutical practice requires also that each
nal container of injection be subjected individually to a physi-
cal inspection, whenever the nature of the container permits,
and that every container whose contents showevidence of con-
tamination with visible foreign material be rejected (1).
The problem with the above USP statement lies with the
word visible, which has the connotation of particles being seen
with the unaided eye. The unaided eye can discern, at best,
particles at sizes of about 4050 m. Detection of smaller par-
ticles cannot be accomplished assuredly with the USPphysical
inspection test. Health care professionals became increasingly
concerned about the aspect of intravenous solutions, espe-
cially large-volume parenterals, contaminated with particles
too small to be seen with the unaided eye, yet still hazardous
when introduced into the veins of a recumbent patient. In the
mid-1970s, the USP and Food and Drug Administration (FDA)
cosponsored the establishment of the National Coordinating
Committee on Large-Volume Parenterals (NCCLVP). The
NCCLVP then established a subcommittee on methods of test-
ing for particulate matter in LVIs. Ultimately, the efforts of
this subcommittee resulted in the establishment of the USP
microscopic assay procedure for the determination of particu-
late matter in LVIs for single-dose infusion and set upper-
limit acceptable particle standards at particle sizes of 10 m
and 25 m (31). These two sizes were also subsequently used
as size standards for particulate matter in small-volume injec-
tions.
208 Chapter 3
The USP standards for LVIs came after standards were
rst established in Australia and Britain. The Australian
standards were based on research by Vessey and Kendall (32);
their results are reported in Table 3.3 along with the upper-
limit particle specications in the British Pharmacopoeia (BP)
and USP.
Because of the widespread acceptance of instrumental
methods for counting and sizing particles, several alternatives
to the present LVI particle limit specications seen in Table
3.3 have been proposed. The National Biological Standards
Laboratory (NBSL) of Australia adopted an approach that de-
pends on the mean and standard deviation of the results from
10 individual containers. This approach takes into account the
usually wide variation in particle counts measured from con-
tainer to container. As seen in Table 3.3, under the Australian
standards, no more than 100 and 2 particles per milliliter at
particle sizes of 5 m and 20 m, respectively, are permitted
Table 3.3 Particulate Matter Standards in Various Compendia
Compared to Those Suggested by Vessey and Kendall
Particle Vessey and British U.S.
size Kendall Australia
a
Pharmacopoeia
b
Pharmacopeia
c
(m) (32) (33) (34) (2)
2 1000 1000 (500)
3.5 250
5 100 100 100 (80)
10 25 50
20 2
25 5
a
Mean count of at least 10 containers using light-blockage method. See text
for additional specications.
b
Particle standards apply only to specic solutions using conductivity
(Coulter Counter). Numbers in parentheses refer to particle limits if light
blockage (HIAC) is used.
c
Particle standards apply only to large-volume parenterals. Particle num-
bers and size determined by microscopic methods unless electronic methods
have been shown to have equivalent reliability.
in LVI solutions. However, the Australian standards also
state that, at the 5-m particle size level, the sum of the mean
and twice the standard deviation is not more than 200; that
is,
x 2s 200
and at the 20 mm size, the sum of the mean particle count
and twice the standard deviation is not more than 4:
x 2s 4
The Australian approach combines the mean values of 10
containers. Hailey et al. (35) suggested the use of a statistical
limit that would account for the mean and standard deviation
of particle counts obtained for each of the 10 containers used.
Their proposal uses a term called statistic S
T
, which is dened
mathematically as
S
T
[(x T)
2
/n]
1/2
where x is the mean value of n results, and T is the target
value (the desired value of the system being measured). For
LVIs, the desired value would be zero (no particles), so the
above equation would become
S
T
[x
2
/n]
1/2
S
T
[S
2
(n 1)/n x
2
]
1/2
where S is the standard deviation. The advantage of the S
T
approach over the Australian draft standard is the increased
stringency of the S
T
requirement on samples near the limit for
mean particle count and having large standard deviation. For
example, if x for 10 containers were 100.0, x 2s were 134.4,
and S
T
were 101.3, the sample would pass the Australian Test
x 2s than 200), but fail the S
T
test (value 100). The disad-
vantage of the S
T
test is that it is based on a target of zero, a
level of cleanliness that can never be achieved. However, to
use any other target value would introduce additional prob-
lems, as discussed by Hailey et al. (35). The authors conclude
by stating that an S
T
target of zero does not add unrealistic
210 Chapter 3
constraints on manufacturers, but it does reduce the risk of
passing a lot of LVI solution having a few containers that are
extremely contaminated with particles.
Groves and Wana (36,37) proposed that a single numeri-
cal value, the index of contamination C, be adopted as a stan-
dard for accepting or rejecting LVI uids. C is obtained from
log particle sizelog particle number plots of solutions mea-
sured by Coulter Counter and light blockage methods in which
the results cross over a size threshold of around 6.0 m. Using
the log-log relationship for the BP limits between 2 m and 5
mand extending the time to 6.0 m, the extrapolated number
of particles equals 63.244. From previous derivations outlined
by Groves and Wana (37), the index of contamination may
then be dened as
C (ln N
1.0
ln 63.244)/m
where N
1.0
is the estimated number of particles per unit vol-
ume at a size threshold of 1.0 m, and m is the slope of the
particle size-number distribution. This index has an advan-
tage over the BP limit tests in that it is not affected by the type
of instrumental principle used to determine the parameters
of the size distribution (see the sections, Electronic Particle
Counters and Comparison of Microscopic and Electronic
Particle-Counting Methods). In addition, C can be calculated
from data acquired at size thresholds that do not necessarily
coincide with those of the BP tests, and calculations of Ccan be
made readily and routinely using on-line computerized quality
control procedures.
The background of the establishment of particulate mat-
ter standards for small-volume injections is reviewed in an-
other later section.
VISUAL INSPECTION: MANUAL METHODS
Manual inspection by human inspectors for the presence of
visible particulate matter in parenteral solutions still remains
the standard, in-line, 100% inspection method. While elec-
tronic television monitors have made signicant strides in re-
placing 100% human inspection, the former remains standard
practice for end-product particle analysis of parenteral prod-
ucts.
Each nal container of a parenteral product must be in-
spected by a trained individual. Any evidence of visible partic-
ulate matter or other product/container defect provides the
grounds for rejecting that container.
Task Group No. 3 of the Parenteral Drug Association pub-
lished guidelines to be considered in the design and evaluation
of visual inspection procedures (38). These guidelines are dis-
cussed in this section.
Equipment
Lighting may be uorescent, incandescent, spot, and/or polar-
ized. The most common source of light is uorescent. The light
source may be positioned above, below, or behind the units
being inspected. The range of light intensity may vary be-
tween 100 and 350 foot-candles.* This intensity can be
achieved either with one 100-watt, inside-frosted incandes-
cent lightbulb or with three 15-watt uorescent bulbs with the
container held 10 inches fromthe light source. Certain types of
products (e.g., colored solutions) or certain types of containers
(e.g., amber) require increased light intensity compared to
that normally used. As light intensity begins to weaken due
to age of the bulbs or usage, lamps should be replaced. Good
practice demands that inspection lamps be monitored periodi-
cally.
A white-and-black background lighted with nonglaring
light is the standard environment used for visual inspection
of product containers. The white background aids in the detec-
tion of dark-colored particles. Light or refractile particles will
appear against the black background.
* The Japanese Pharmacopeia (JP) reqires 740930 foot-candle light intensity for
inspection of injections contained in plastic containers (39).
212 Chapter 3
Fig. 3.1 Apparatus for visible particle inspection as presented in
the European Pharmacopeia, Figure 2.9.20.-1, p. 222.
The EP provides a gure of the type of apparatus to be
used in visible inspection of particulates (3) (Fig. 3.1). The ap-
paratus consists of
a matte black panel of appropriate size held in a vertical posi-
tion
a nonglare white panel of appropriate size held in a vertical
position next to the black panel
an adjustable lampholder tted with a suitable, shaded,
white-light source, a suitable light diffuser (details pro-
vided in the EP), and illumination intensity maintained
between 2000 lux and 3750 lux, with higher intensities
preferred for colored glass and plastic containers
Many manufacturers have progressed to the use of auto-
mated inspection equipment. Appropriate inspection proce-
dures should specify and monitor the adjustment of the ma-
chines operating parameters required to achieve a quality
inspection that is at least equivalent to that resulting from a
previously established manual inspection procedure. Auto-
mated inspection machinery is discussed in the section, Vi-
sual Inspection: Automatic Methods.
Astandard inspection booth contains an all-black interior
except for the front entrance for the inspector. A vertical
screen in the back of the booth is half black and half white.
Light usually is projected vertically with frontal blockage to
protect the observers eyes from direct illumination. A magni-
fying lens at 2.5 magnication may be set at eye level to aid
the inspector in viewing the container in front of the white/
black background. Excellent viewing is provided without dis-
traction, and acuteness of vision is increased to improve the
level of discrimination. It could be argued that the level of dis-
crimination becomes too high; that is, containers are rejected
that would not have been rejected had no magnication been
used.
Inspection cabinets should have black side walls with a
bafe to prevent the light source fromimpinging on the inspec-
tors eye. Fluorescent lamps provide a better light source be-
cause they are more diffuse than incandescent lamps.
Methodology
Most inspection processes are referred to as off-line inspec-
tions, in which the inspection procedure occurs at the comple-
tion of the manufacturing, lling, and sealing process. In-line
inspection of container components can also be done, espe-
cially if the production process can be suitably adapted to
achieve the desired results without increasing the risk of mi-
crobial and particulate contamination. Obviously, the removal
of defective containers, such as those showing cracks or the
presence of particles, prior to the lling of the product ensures
product quality and minimizes loss of expensive drug prod-
ucts.
Standard operating procedures for inspection of paren-
teral containers depend on the kind of container inspected;
214 Chapter 3
Table 3.4 Basic Procedure for Manually Inspecting Clear
Solutions for Visible Evidence of Particulate Matter
1. Container of parenteral solution must be free of attached
labels and thoroughly cleaned. Use a dampened, nonlinting
cloth or sponge to remove external particles.
2. Hold container by its top and carefully swirl contents by
rotating the wrist to start contents of the container moving in
a circular motion. Vigorous swirling will create air bubbles,
which should be avoided. Air bubbles will rise to the surface
of the liquid; this helps to differentiate them from particular
matter.
3. Hold the container horizontally about 4 inches below the light
source against a white-and-black background. Light should be
directed away from the eyes of the inspector, and hands
should be kept from under the light source to prevent glare.
4. If no particles are seen, invert the container slowly and
observe for heavy particles that may not have been suspended
by swirling.
5. Observation should last for about 5 seconds each of the black-
and-white background (24).
6. Reject any container having visible particles at any time
during the inspection process.
that is, procedures will be slightly different for ampules than
for large-volume glass bottles, for amber vials than for int
vials, and for plastic bags than for glass containers. However,
a basic procedure can be followed regardless of the type or size
of container; an example of such a procedure is given in Table
3.4.
Personnel
The human inspector determines the quality and success of
the manual inspection process. Since the inspection process is
subjective in nature, the main limitation of the process lies
with restriction in the vision, attitude, and training of the indi-
vidual inspector.
As a minimum standard, personnel assigned as inspec-
tors should have good vision, corrected, if necessary, to ac-
ceptble standards. Inspectors should not be color-blind. Visual
acuity should be tested at least on an annual basis.
Good attitude and concentration cannot be overempha-
sized. One of the major limitations of human inspection for
particulate matter is reduced efciency of the individual be-
cause of a lack of concentration. This can easily occur if the
inspector suffers fromextreme worry or other distraction from
outside personal pressures. Obviously, emotional stability is
an important criterion in selecting inspectors.
Fatigue also becomes a major limitation of human inspec-
tion. Personnel should be provided appropriate relief from the
inspection function by rotating jobs and allowing for rest pe-
riods.
Formal training programs must precede the acceptance
of an individual as a qualied inspector. The training program
should include samples of both acceptable and unacceptable
product containers that must be distinguished by the trainee.
During the training period, all units inspected by the trainee
should be reinspected by qualied inspectors to ensure the
quality of the inspection and the development of the trainee.
After the inspector has passed the training period, perfor-
mance tests should be done at random intervals to ensure that
quality standards are being maintained.
Two reports have been published concerning the effect of
personnel experience on detection of particles in ampules.
Graham et al. (40) found that inspectors with no experience
and inspectors having at least 10 years experience agreed 64%
to 83% of the time regarding ampules inspected under various
conditions. Experienced inspectors were faster in the inspec-
tion process. Baldwin et al. (41) found that experienced inspec-
tors reject ampules at a greater rate (28.3%) than did nonexpe-
rienced inspectors (13.2%). Discrimination in particle
detection apparently correlates with training and experience.
Acceptance Standards
Visible evidence of particulate matter in parenteral products,
both solids and liquids, is considered by most parenteral prod-
216 Chapter 3
uct manufacturers to be a very serious (critical) defect. There-
fore, acceptable quality levels (AQLs, the highest percentage
of defective units unacceptable for releasing the batch) for sta-
tistical sorting samples taken for particulate and other quality
inspection generally are within the range of 0.25% to 1.0%.
For example, with a sample of 315 vials and an AQL of 0.25%,
nding 2 vials with visible foreign particulates will be accept-
able, but nding 3 vials with particulates will cause rejection
or resorting of the batch.
Japanese Method for Inspection and Analysis
of Particulate Matter
Requirements for freedom of parenteral solutions from the
presence of particulate matter are very strict in Japan. Inspec-
tion of individual containers for any visible evidence of partic-
ulate matter is done much more rigorously. For example, an
inspector in a typical Japanese pharmaceutical company will
take up to 10 seconds inspecting a single vial of a parenteral
solution. Contrast this with inspectors in a typical fast-speed
American parenteral manufacturer; they will inspect 50150
vials per minute for evidence of particulate matter.
The Japanese technique for preparation and testing of so-
lutions for the presence of particulate matter by microscopic
analysis deserves some attention. The meticulousness of their
preparation techniques are impressive. For example, all mate-
rials (forceps, petri dishes, ltration funnels) used in ltering
solutions are rst sonicated for at least 5 minutes, then
washed thoroughly with particle-free water three times. The
membrane lters used for the blank controls are washed thor-
oughly using a very rigid procedure involving starting at the
top of the nongridded side of the lter, sweeping a stream of
particle-free water back and forth from top to bottom, then
repeating this on the gridded side of the lter. After inserting
the lter into the lter holder base and installing the funnel,
the entire system is rinsed twice with particle-free water, tak-
ing care not to allow the rinsings to pass through the lter.
Further rinsings are completed with the water vacuumed
through the lter. Interestingly, this water is introduced into
the funnel using an injection syringe tted with a 0.45-m l-
ter. The maximumallowable number of particles for the entire
membrane lter pad used as the blank control is 3 that are
10 m or longer and 1 that is 50 m or larger using a suitable
microscope with 40 and 100 magnication with incident
light at an angle of 20.
Sample test solutions are handled in the same way. Five
vials are ltered through the same lter pad. Some Japanese
companies even use a lter pad that is only 4 mm in diameter.
Filter pads, after vial contents have been ltered through the
pads, are photomicrographed, usually at the 40 magnica-
tion. Test results are judged by visual comparison of the test
lter pads with reference photographs of previous test sam-
ples judged by the Quality Control Department to represent
the particulate quality desirable with the product sample.
Comparison to Other Particle Inspection Methods
Manual visual inspection often is criticized because of its ap-
parent inconsistency and unreliability. Its subjective nature,
depending on the judgment of uncontrolled and variable hu-
man evaluation of what may or may not be seen, drives many
quality control specialists to seek other methods for achieving
the same purpose100% nondestructive inspection of paren-
teral products for the presence of particulate matter.
Manual inspection can be compared to automatic elec-
tronic inspection methods on the basis of precision and accu-
racy (42). Precision is related to consistency, which measures
the capability of any given process to detect the same condi-
tions in repeated blind tests. Accuracy relates to bias, based
on inequality of reject rates. When the precision and accuracy
of manual inspection were compared to those demonstrated
by Autoskan, an electronic video particulate inspection ma-
chine (see the section, Autoskan System), two interesting
conclusions were drawn:
1. Consistency of both methods, based on Cohens kappa sta-
tistic, was comparable. Autoskan was not superior to man-
218 Chapter 3
ual inspection in terms of repeatedly rejecting those am-
pules containing particles.
2. Based on Cochrans Q statistic, bias was a problem with
human inspection, while appropriate settings of the Au-
toskan could eliminate machine bias. However, rejection
rates were established using only one machine, while
eight inspectors were tested and compared for their rejec-
tion rates for the same batch of ampule products. An ex-
ample of rejection rates comparing machine and human
inspectors is given in Table 3.5.
Reproducibility in human visual inspection was the sub-
ject of an article by Faesen (43). Each of 1000 diamond-num-
bered ampules and vials was inspected by 10 different inspec-
tors, twice for each operator, using a Liquid Viewer, an
inspection cabinet, and a Rota Ampul Inspection Machine (for
ampules only). The total number of rejects was registered fol-
lowing 60 inspections for the 1000 ampules (three inspection
methods performed twice) and 40 inspections for the 1000
vials (two inspection methods performed twice). Results indi-
cated that reproducibility in a visual inspection was nearly
twice as high when performed with a Liquid Viewer as com-
pared with reproducibility for inspections performed with the
inspection cabinet. For ampules, the value reached using the
Ampul Inspection Machine was less than 10% units higher
Table 3.5 Reject Rates (%) for Four Products
Inspected for Particulate Matter by Autoskan
and Eight Human Inspectors
Eight human inspectors
Product Autoskan Average Range
A 22.2 25.2 19.729.2
B 20.2 21.5 17.524.0
C 20.3 17.4 15.519.5
D 7.2 7.9 5.610.0
Source: Ref. 42.
than that with the inspection cabinet. The Liquid Viewer ap-
peared to be the superior instrument for visually inspecting
parenteral solutions.
In a panel discussion of mechanical inspection of ampules
(4446), it was stated that the average reject rate for ampules
inspected manually was 22.5%. Three inspection machines
were compared to manual inspection. Autoskan equipment
(44) showed a reject rate of 14.1%, the Rota Machine (45) in-
creased the reject rate 0.51.5% over that for manual inspec-
tion, and the Strunck Machine (46) yielded a reject rate of
2.98%. The major advantage of machine inspection was simply
the substantial increase in the number of containers inspected
per unit time.
Blanchard et al. (47) compared the human visual exami-
nation method with several other methods for detection of par-
ticulate matter in large-volume parenteral solutions. Visual
methods using either the naked eye under diffuse light or a
2.5 opter lens under diffuse light proved to be inadequate to
other methods (light scattering, Prototron, and microscopic ex-
amination after ltration) in terms of sensitivity to low levels
of particulate contamination. Not surprisingly, the visual
method showed a high degree of subjectivity.
Visible Particle Sizes to the Unaided Eye
Since the number and size of particles in parenteral solutions
have become important characteristics to evaluate, it has been
assumed that particles larger than 40 or 50 m are detectable
by the unaided eye. Thus, in complying with USP require-
ments that any container showing visible evidence of particu-
late matter be rejected, it must be assumed that the average
inspector will pass those solutions containing particles with a
size of 40 m or less. This, of course, presents some discomfort
for those who believe that particulate matter, especially in the
size range of 1040 m, is clinically hazardous.
It is not only the size, but also, and probably more impor-
tantly, the number of large particles injected into man intrave-
220 Chapter 3
nously that is considered dangerous. Thus, ofcial standards
have been enforced for maximum allowable numbers of cer-
tain size particles in parenteral solutions.
At least one attempt has been made to quantify the size
and concentration of particles that can be detected by the un-
aided eye (8). In a standard booth, 5 ml ampules containing
10 to 500 particles per milliliter of particle sizes between 5
and 40 m (using polystyrene beads) were inspected by 17 in-
spectors. Based on a multiple linear analysis model that calcu-
lated the probability of rejecting an ampule as a function of
particle size and concentration, sizes of particles detected at
various concentration levels at 50% and 100% probability of
rejection rates were predicted. These data are reproduced in
Table 3.6. The authors concluded that a 50% probability of
rejection rate be achieved with 20-mparticles in sample solu-
tions in order for potential inspectors to be qualied for in-line
inspection. However, it is interesting to note that a minimum
Table 3.6 Size of Particles of Varying Probability Levels
a
Particle size (m), Particle size (m),
Particle concentration 50% chance 100% chance
USP limit 50 particles/ml
b
18.82 51.45
USP limit 5 particles/ml
c
19.96 54.88
1-ml ampule, 1 particle 20.07 55.21
50-ml vial, 1 particle 20.10 55.29
1-liter large volume, 20.10 55.29
1 particle
Source: Ref. 8.
a
Arcsin P
1
0.33689252 0.02231515 size 0.000035 size versus concen-
tration 0.00008694 concentration.
b
Not more than 50 particles/ml equal to or larger than 10 m.
c
Not more than 5 particles/ml equal to or larger than 10 m.
particle size of 55 m was required for all inspectors to reject
all solutions containing this size of particle.
VISUAL INSPECTION: AUTOMATIC METHODS
Introduction
Manual visual inspection continues to be the most commonly
used quality control method for particle detection in paren-
teral products. The limitations of depending on human inspec-
tion for rejecting particle-contaminated solutions have al-
ready been addressed. High technology strives for
sophisticated automatic methodology to replace the depen-
dency on human manual inspection. One area of high-technol-
ogy application to particle analysis in parenteral products is
the development and improvement of electronic particle
counters. The main limitation in the use of these instruments
in particulate matter analysis resides in the fact that the tests
are destructive. For each nal container of parenteral product,
100% inspection cannot be accomplished with electronic parti-
cle counters. The same limitation holds true for automated mi-
croscopic methods. The area of technology that offers the
greatest potential for replacing human examination in 100%
container inspection requirements is the area of computer-
controlled, automatic electro-optic systems. Such systems are
rapid, nondestructive, and reproducible in their inspection of
parenteral products for foreign matter.
Early attempts to automate the 100% inspection process
were reviewed by Groves (24). Systems developed and tested
included the Brevetti device (48), Strunck machine (49), and
the RCAmachine (50). Despite considerable electronic ingenu-
ity, all of these systems required human intervention at some
stage of the inspection process, although Groves admitted this
may be a consolation.
Hamlin et al. (51) were among the rst investigators to
test the use of television (TV) as an inspection device in de-
tecting subvisible particulate matter. However, their main
emphasis in using TV monitoring was as a research tool in
222 Chapter 3
detecting 10-m particles in experimental formulations for
prediction of estimated shelf life based on physical stability.
Also, TV monitoring required human involvement in viewing
and rejecting particle-contaminated solutions.
Technology has made signicant improvements in fully
automated parenteral product inspection procedures. Disad-
vantages of earlier automated systems, such as lack of stan-
dardization of performance, separating marks on the outer
container surface from particles inside, failures to detect un-
derlls or empty containers, and machine variabilities, have
largely been eliminated with the automated systems available
today.
Video inspection employs one of two basic mechanisms
for automated container inspection (52). One mechanism uses
imaging optics in which the particles suspended in the solu-
tion are illuminated by a ber-optic light system and imaged
on a video display. The other mechanism employed in auto-
mated video inspection is based on light scattering from par-
ticulate matter, which is then received by a detection system
and projected onto a television camera system. Several sys-
tems commercially available employ the light-scattering prin-
ciple for automated video inspection. Among the most widely
used systems are the Autoskan system, the Eisai Ampul In-
spection Machine (AIM) system, Particulate Detection System
Ampules and Vials and the Schering PDS/A-V system. The
Prototron system (53,54) at one time was a widely used nonde-
structive inspection method using laser light, but it is no
longer used today.
Autoskan System
The Autoskan system uses white lightin contrast to laser
light, which was used by the Prototron systemto illuminate
particles suspended inparenteral solutions. Particles will scat-
ter the light, which is received by a television camera system.
Any solution that contains particles will generate an error sig-
nal, and that product container will not be released by the Au-
toskan system at the accept station. Containers are also auto-
matically rejected if they are either underlled or overlled.
Autoskan became the rst totally automatic inspection
system developed to detect particulate matter in injectable so-
lutions. The instrument is suitable for the inspection of vials,
ampules, cartridges, and syringes. In Fig. 3.2, ampules on a
rotary feed table are fed into the turret. The turret picks up
the ampules and intermittently transports themaround to the
inspection station, where the lens of the television camera is
located. The ampule is magnied by high-intensity light from
belowthe check holding the ampule (see Figure 3.3). This light
reects particles moving in the liquid, making them visible to
the camera (often also visible to the human eye). The Au-
toskan checks in the turret contain motors that spin each con-
Fig. 3.2 Autoskan inspection system showing ampules on a rotary
feed table leading into and exiting fromthe detection area. (Courtesy
Lasko Company, Leominister, Pennsylvania.)
224 Chapter 3
Fig. 3.3 Close-up of ampule being inspected for particles by the
Autoskan system. (Courtesy Lasko Company, Leominster, Pennsyl-
vania.)
tainer at an adjustable speed until the container comes before
the lens of the television camera. The spinning is designed to
dislodge and set particles in motion and create a central vortex
in the liquid. This permits the television and electronic system
to detect underlled, overlled, or empty containers. The in-
spection area of the container is preset. Liquid levels that do
not exactly t within the upper and lower limits of the inspec-
tion area are rejected automatically. If the container has the
correct ll volume, it then becomes eligible for the inspection
process that detects the presence of foreign matter.
A master picture of the correctly lled container is
taken simultaneously with the liquid level pictures. The mas-
ter is put into Autoskans electronic memory, which serves as
a standard for subsequent comparative video images of the
same container. Sixteen comparison pictures of the container
are taken and compared to the master picture. Any difference
between the master and any of the subsequent comparitive
pictures of the single container will result in that container
being rejected. Since the Autoskan has the capability of in-
specting between 1800 and 4500 containers per hour, the time
for checking the liquid level, taking a master picture, and sub-
sequent comparison pictures is less than 1 second per con-
tainer.
The fact that the liquid contents are swirling while the
container itself is motionless during the inspection process has
a very important implication. The master picture is based on
a motionless container. All scratches, printing, or other marks
on either the outer or the inner surface of the container are
part of the master picture. Any difference between the master
and any one of the subsequent comparison pictures of the sin-
gle container, therefore, would be caused only by particulate
matter moving within the liquid contents, reecting light back
to the camera.
Louer et al. (55) compared Autoskans performance
against visual inspection for discrimination of good and
bad ampule solutions in terms of particulate contamination.
Their results, reproduced in Table 3.7, showed that, on the
average, there was a 93% rejection of control bad ampules by
the machine, whereas the percentage for visual inspection was
Table 3.7 Comparison of Visual Inspection Versus Autoskan
Automatic Inspection of Ampules for Particulate Matter
Visual inspection, Machine
50 examinations inspection,
(average) 10 runs
Ampules (n 50) rejected from 26.6 42.7 of 46
the control bad lot
Percentage of bad lot rejected 53.2 92.8
Ampules rejected from the con- 3.16 39.6 of 752
trol good lot
Percentage of good lot rejected 0.42 5.25
Source: Ref. 55.
226 Chapter 3
only 54%. The rejection of good ampules by visual inspection
was signicantly lower than the rejection of good ampules by
the machine.
Eisai Ampul Inspection Machine System
Like the Autoskan system, the Eisai AIM system uses white
light as the source of detection of particles. However, whereas
Autoskan measures light scattered from a particle, Eisai de-
tects the moving shadows produced by foreign matter in a con-
tainer of solution. As with the Autoskan, each container is
spun around and stopped so that only the liquid in the con-
tainer is still rotating when the container enters the light. If
any foreign matter is oating and rotating in the liquid, the
light transmitted through the liquid is blocked, and a shadow
is cast by the moving particles. Eisai systems employ a photo-
transistor that converts moving shadows into electrical sig-
nals. These signals are compared to preset detection sensitiv-
ity signal standards, and if the standard sensitivity is
exceeded, the container is rejected. Like the Autoskan, the Ei-
sai detector does not react to scratches, stains, colors of the
ampule, or the color of the liquid contents since these are all
perceived as stationary objects.
The Eisai system, like the Autoskan system, checks the
volume of liquid in the container and can reject overlled, un-
derlled, and empty containers. The shadow cast by the liquid
meniscus of a properly lled container is expected to fall
within a certain preset range within the inspection eld. If
it falls above or below this range, the container is rejected.
Adjustments in the Eisai system can be easily made for differ-
ent ampule sizes, ampule color, and viscosity of the liquid con-
tents.
The conveyance-and-inspection mechanism of the Eisai
system is shown in Fig. 3.4. Ampules are conveyed by the star
wheel onto the inspection table, spun at a high speed, and
stopped before reaching the light beam. When the ampule en-
ters the light beam, the light projector and detector follow the
Fig. 3.4 Conveyance and inspection mechanism of the EISAI auto-
matic inspection system. (Courtesy Eisai, USA, Inc., Torrance, Cali-
fornia.)
ampule while liquid is still rotating inside. After one ampule
is inspected by two sets of projectors and receptors (thus, a
double-inspection system), the next ampule is carried through
the same process. Ampules are moved by the screw conveyor
to the sorting pendulum, where rejected and accepted ampules
are separated. The AIM system automatically keeps count of
the number of accepted and rejected containers and displays
these numbers on the display panel.
Performance evaluations of the Eisai AIM system have
been conducted by at least three major pharmaceutical compa-
nies: Pharmacia (now Pzer), Organon, and Merck. The com-
plete report of these evaluations is available from Eisai USA,
Incorporated. One evaluation was reported at the 1982 annual
meeting of the Parenteral Drug Association. All three investi-
228 Chapter 3
gators concluded that the inspection of ampules by the AIM
system compared to manual inspection resulted in a great im-
provement in the quality of ampules accepted and released.
Using the performance criteria model published by Knapp et
al. (discussed in the section, Probabilistic Particulate Detec-
tion Model), Upjohn found that the Eisai machine will do a
better job than manual inspection in rejecting defective am-
pules for production lots. Organon found the Eisai machine to
be more reproducible than human inspection as long as the
product in the containers was not an oil or had the tendency
to foam. Merck found that the quality of the Eisai-inspected
material was more than twice as good as the quality of those
containers manually inspected. Advantages of the Eisai sys-
tem itemized by the Merck report include (1) versatility, that
is, ability to handle a large variety of ampule and vial sizes,
products, and viscosities; (2) adjustable sensitivity level; (3)
attainable speeds; (4) results of performance studies; and (5)
price.
To our knowledge, there is no published report directly
comparing the inspection performances of Autoskan and
Eisai.
Schering PDS/A-V System
Schering Corporation has patented the PDS/A-V, a fully auto-
mated particulate inspection system (56). A photograph of the
system is shown in Fig. 3.5.
Containers are conveyor fed from oriented trays into the
inspection star wheel. Light is directed into a container using
fused ber-optic pipes formed into a narrowslit. The container
is spun, creating motion of particles in the liquid inside. The
entire container is scanned by a ber-optic image dissector,
which forms multiple-image planes of the entire liquid vol-
ume. The image dissector transmits light scattered from mov-
ing particles in the container to a set of matched photodiodes,
where the light is changed into an electrical signal and pro-
cessed. Only signals from moving particles are processed;
Fig. 3.5 Particulate Detection System 100 for ampules and vials.
(Courtesy Electro-Nucleonics, Inc., Faireld, New Jersey.)
thus, container defects or printing do not generate false re-
jects. The image dissector inspects rst the lighted lower part
of the container for glass particles, then the full volume of the
container for other particles, including those oating at the
meniscus. Containers are rejected by a single-board microcom-
puter if the scattered light detected results in a higher score
than the digital rejection criteria stored in the computer. The
device can inspect 10,000 containers per hour. More elaborate
230 Chapter 3
details of the successful automated inspection device are given
in articles published by Knapp et al. (5661).
Probabilistic Particulate Detection Model
Knapp and coworkers published a series of articles describing
the theory and application of a probabilistic inspection model
in the automated nondescriptive particulate analysis of sealed
parental containers. In his most recent publication (57),
Knapp compares the application of his probabilistic model and
statistically dened particle contamination quality regions to
the application of the probabilistic concepts developed for heat
sterilization validation (e.g., Fo) with respect to improving
product quality.
The probabilistic model is based on the nding that par-
ticulate inspection methodologies, human or robotic, are prob-
abilistic rather than deterministic in nature (58). In other
words, no nal container of solution is acceptable or unaccept-
able; rather, each nal container of solution possesses a proba-
bility of being rejected for whatever inspection process is being
evaluated. Rejection probabilities are determined simply by
recording the number of times a numbered container is passed
and the number of times that same container is rejected dur-
ing a manual or automatic inspection process. Each container
accumulates an accept/reject record. If 1000 containers are in-
spected several times and each of the 1000 containers yields
an accept/reject ratio, a histogram can be constructed plotting
the number of containers in each probability group against an
empirically determined rejection probability. Such a histo-
gram is shown in Fig. 3.6, and it represents the cornerstone
for the conversion by Knapp et al. of particulate inspection
from a craft to a science (61).
The abscissa in Fig. 3.6 represents rejection probabilities
grouped arbitrarily into 11 intervals. The ordinate represents
the logarithmic number of containers (vials) within each of the
11 probability groups. For example, of the 1000 vials inspected
for particulate contamination, 805 vials were found to be par-
Fig. 3.6 Histogram plotting number of vials per each probability
of rejection group. (Courtesy J. Z. Knapp, Schering Corporation,
Kenilworth, New Jersey.)
ticulated free in each of the 50 inspections, while 2 vials con-
tained particulates that were detected in each of the 50 inspec-
tions.
The dashed lines on the lower half of the histogram show
the average number of vials rejected in a single inspection or
two sequential inspections in each probability group. These
values are obtained from the relationship (58)
P(Mn)
i
P(MI)
i
232 Chapter 3
where P(Mn)
i
is the rejection probability associated with the
nth manual inspection in a probability group, P(MI)
i
is the
quality of vials rejected in a rejection probability group in a
single inspection, and n is the number of inspections of re-
jected material. For example, of the eight vials located in the
0.6 rejection probability P(MI)
i
group, ve were rejected fol-
lowing a single inspection, while only three were rejected fol-
lowing two sequential inspections. This indicates that im-
proved discrimination occurs following a reinspection of initial
rejects. The reinspection was utilized as a practical response
to the existence of particulates even in well-controlled paren-
teral manufacturing areas below the range of present medical
and FDA interest (59). From the information contained in the
reinspection histogram of Fig. 3.6, Knapp and Kushner (58)
dened three zones within the rejection limits of 0 and 1.
The accept zone contains all vials that have less than 1
chance in 10 of rejection in two sequential inspections. The
reject zones contain all vials that have at least 1 chance in 2
of being rejected in two sequential inspections. The gray zone
exists between the accept and reject zones. For single inspec-
tions, the probability limits for the three zones are seen in Fig.
3.6, where
Accept zone, p .3
Gray zone, .3 p .7
Reject zone, p .7
Figure 3.6 also shows three terms: RZN, RZR (M1), and
RZR (M2). The denitions of these terms are given in the g-
ure. Their calculations are explained thoroughly in Ref. 58.
Using these terms, a variety of parameters can be measured,
including reject zone efciency (RZE) and undesired reject
rate (RAG). By denition, RZE RZR/RZN. In the example
in Fig. 3.6, the RZE after a single inspection is 81.7%. This
means an 81.7% probability exists for a manual single-inspec-
tion method to reject those vials known to exist in the reject
zone. Matching or exceeding this objective measure of the se-
curity achieved by a manual parenteral inspection procedure
should be the only good manufacturing practice (GMP) re-
quirement for validation of any alternative inspection tech-
nique or process (58).
The availability of the probabilistic model for particle in-
spection of sterile product solutions in their containers has
permitted objective evaluations of various inspection parame-
ters, new methodologies, and new detection equipment. For
example, the Schering Particulate Detection System for
Ampuls (PDS/A) was validated using the probabilistic meth-
odology (56,59). RZE scores were used to determine the effects
of lighting levels, light polarization, and lens magnication on
a human inspection of vials mechanically positioned by an ex-
perimental machine at Upjohn (now Pzer) (62). RZE scores
permitted the selection of optimal settings for light, magni-
cation, and light polarization. Interestingly, however, RZE
scores also showed that the mechanical handler was not as
efcient in meeting the minimum Upjohn standards for per-
formance as their currently used inspection process. The prob-
abilistic model allowed a valid decision to be made based on
objective scientic data.
Knapp concludes his most recent publication (57) by stat-
ing that this methodology is now used worldwide and is being
applied to dene the performance of most automated particle
inspection systems.
CURRENT ISSUES WITH VISIBLE PARTICLE DETECTION
AND INSPECTION PROCEDURES
Inspection of solutions for visible particulates has taken on
newpriority and scrutiny in the past several years. Part of this
is because of increased FDA and other regulatory compliance
groups demanding higher degrees of product quality with re-
spect to visible particulates. Another part is the struggle of
the industry to dene quality policies related to visible partic-
ulates in light of the dichotomy that, although everyone ac-
cepts the fact that particulates can and do exist in solutions
(i.e., you will never see a solution with zero particulates), we
234 Chapter 3
are supposed to approve and release only those products that
supposedly have no particulates. Where do you draw the line
with respect to acceptable or unacceptable product quality?
There are signicant struggles with interpreting the
pharmacopeias with respect to visible particle testing for par-
enterals.
All sterile products are subjected to a 100% inspection,
typically automated to some degree. However, each unit of
product is viewed/analyzed by human inspection for only a
second or two (at most maybe 5 seconds, with the exception
of Japan, which may inspect for up to 10 seconds per unit of
product). Following 100% inspection, product lots are sub-
jected to a more demanding release test, described in the com-
pendia. In fact, the compendia have the following denitions:
USP: essentially free from particles that can be observed on
visual inspection
EP: clear and practically free from particles
JP: clear and free from readily detectable foreign insoluble
matter
These compendial requirements plus current industry prac-
tices with respect to visual inspection beg the following ques-
tions (63):
1. How are 100% on-line inspections separated from release
testing when on-line inspections require rejection of all
product units that contain particles while release testing
requires that product units be essentially free from parti-
cles?
2. How are the compendial requirements essentially free
from particles to be interpreted?
3. Is there any possibility that inspection methodology for
visual particulates will be hormonized worldwide?
4. Is there any possibility that haze testing can be harmo-
nized and acceptable to all compendia?
FDA presentations have indicated that quality control of
visible particulate matter is of great concern to inspectors. A
review of Warning Letters issued in 1999 and 2000 (64) from
the FDA to manufacturers of injectable products provides the
most often cited specic observations with respect to concerns
about visible particulates:
1. For the same lot, there is sometimes a high rejection rate
by certain personnel, while other personnel have low re-
jection rates.
2. The results from statistical sampling at the beginning,
middle, and end of a lling operation found no defects,
whereas the entire lot was rejected for release because of
visual defects during 100% inspection.
3. Failure to establish a maximum acceptable level of vials
rejected during 100% inspection.
4. Failure to take adequate action to ensure the quality of
released product when the particulate reject limit was ex-
ceeded.
Personnel responsible for detection of visible particulate
matter must be thoroughly trained for this important quality
evaluation. Training is not an easy task because of a variety
of reasons: vision capabilities, concentration, sample stan-
dards of particulate types, inspection environment, and quali-
cations of the trainer(s).
USP TEST FOR SUBVISIBLE PARTICULATE MATTER
IN LARGE-VOLUME INJECTIONS FOR SINGLE-DOSE
INFUSION AND SMALL-VOLUME INJECTIONS
After several years of collaborative effort among laboratories
from the FDA, universities, and pharmaceutical manufactur-
ers, a method became ofcial in the First Supplement of the
USP (19th edition) in 1975 for the particulate matter analysis
and release specications for single-dose large-volume paren-
terals (LVP now LVI). The original method involved the l-
tration of 25 ml of solution through an ultraclean membrane
ltration assembly, then observing the membrane and count-
ing entrapped particles on its surface under a microscope us-
ing 100 magnication.
236 Chapter 3
Since the advent of the microscopic test for LVI solutions,
the particle load in these solutions was substantially reduced.
This was recognized in the early 1980s such that attention
turned to establishing subvisible particle standards for small-
volume injections (SVIs). By the USP XXI (July 1985), the
electronic light obscuration (LO) test method was introduced
for SVIs. Until around 2000, LVIs were evaluated for subvisi-
ble particulate matter using the microscopic method, while
SVIs were primarily evaluated by the light obscuration
method. However, effective with the USP XXV, both LVI and
SVI solutions are now primarily evaluated for the presence of
subvisible particulates by the light obscuration method. If the
light obscuration method cannot be used, then the microscopic
method is allowed. Many companies perform both methods in
measuring the amounts and types of particulate matter in
their parenteral solutions and reconstituted powders. Also, ef-
fective with the USP XXV, both LVIs and SVIs followed the
same general USP guidelines (test apparatus, calibration, test
environment, test procedures, and calculations) for both the
light obscuration particle count test (pp. 20462052) and the
microscopic particle count test (pp. 20462052).
Development of the Small-Volume Injection Subvisible
Particle Test
Back in the early 1980s, there was a general consensus that
these LVI limits were too strict for SVI solutions and, in fact,
SVIs should not have particle limits because (a) volumes ad-
ministered are much smaller than those for LVIs, and (b)
health hazards from injected particulates were not unequivo-
cally established. Nevertheless, the USP sponsored studies to
establish particle limits for SVI solutions reasonable for both
a safety standpoint and a quality control standpoint achiev-
able by the parenteral industry.
Two SVI particle limit proposals were published in late
1983 (65). One was based on particles per container proposed
by the USP Subcommittee on Parenteral Products. The other
proposal, by the USP Panel on Sterile Products, was based on
particles per milliliter. A comparison of the two proposals is
summarized below:
25 m 10 m
Subcommittee pro- 1000 Particles per 10,000 Particles per
posal container container
Panel on sterile 70 Particles per 250 Particles per
products proposal milliliter milliliter
The subcommittee proposed limits based on the following
rationale: The addition of up to ve containers of any SVI to
a 1-liter LVI solution should not increase the number of parti-
cles by more than double those allowed by the USP limit for
LVI solutions (5 particles per milliliter 25 m or 5000 parti-
cles in 1 liter 25 m; 50 particles per milliliter 10 m
or 50,000 particles in 1 liter 10 m). If ve additives, each
containing no more than 1000 particles per container 25 m,
were admixed with the 1 liter LVI containing 5000 particles
25 m, the total particles 25 m or larger would be 10,000,
which would be the maximum allowable particle number per
admixed solution. At 10 m, the total particle number with
ve additives in a 1-liter LVI would be 100,000, which would
be no more than double that of the LVI alone. Therefore, the
subcommittee proposal was based on concern more for the cu-
mulative particulate insult the patient might receive than for
the number of particles per milliliter of solution.
The USP panel proposal was based on data from an FDA
survey of a large number of small-volume parenterals. Particle
counts were obtained from an electronic particle counter
(HIAC, High Accuracy Instruments Division). After combin-
ing data from all the samples (157 samples of 19 aqueous drug
products, 10 units per each sample), the upper 95% condence
limits of the means at 25 m and at 10 m were those listed
above.
238 Chapter 3
At that time, of the more than 500 ofcial SVI products
in the USP, 134 of these products (65) meet the following crite-
ria for selection of products subject to particulate matter limit:
1. The drug is usually administered via the artery or vein or
intrathecally.
2. The drug is likely to be used continuously or repeatedly
for a course of treatment.
3. Drugs solely for emergency use, for diagnostic procedures,
for anticancer therapy, or for episodic use are excluded.
While many laboratories preferred to employ the LVI mi-
croscopic method for counting particles in SVI products, the
USP XXI introduced the use of an electronic liquid-borne par-
ticle counter system. Initial controversy over the test resulted
in a postponement of the test becoming ofcial until July 1985.
The major complaint of the new USP method centered around
the use of the HIAC-Royco electronic particle counter. Like
any electronic counting device, the HIAC cannot identify and
characterize particles, cannot accurately measure a particles
longest dimension (i.e., it measures all particles as spheres),
will count silicon and air bubbles as particles, and
standardization/calibration of the HIAC can be difcult. Also,
many manufacturers objected to being forced to use an instru-
ment that at that time was available from only one major U.S.
supplier.
Other concerns over the proposed USP test for SVI partic-
ulate matter included lack of a sufcient database from which
limits were established; lack of validation of the USP proposed
method; the basis for requiring particle limits for some prod-
ucts but not for others in individual monographs; problems
with specic details in the calibration, preparation, and deter-
mination sections of the test; and the lack of consistency be-
tween the LVI and SVI tests for particulate matter. These are
discussed in more detail. The USP (Section 788) requirement
for particulate matter in SVIs became effective in 1986. The
test called for the use of an electronic liquid-borne particle
counter system utilizing a light obscurationbased sensor
with a suitable sample-feeding device. The USP recommended
Pacic Scientic, makers of HIAC/Royco particle counters,
and using sensors manufactured by Russell Laboratories.
ELECTRONIC PARTICLE COUNTERS
The limitations of human inspection and microscopic analyti-
cal methods in the detection of particulate matter in injectable
products have necessitated the use and advancement of elec-
tronic particle-counting methods in the pharmaceutical indus-
try. In 1986, the USP adopted for the rst time an electronic
particle-counting method to be used in particulate matter test-
ing of small-volume injections. Much controversy over the
type, standardization, and limitations of electronic particle-
counting methods has continued over the years. The 1987 In-
ternational Conference on Particle Detection, Metrology, and
Control concluded with the general perception that there re-
mained many measurement problems with electronic particle
counters (66). However, the 1990 Particle Conference closed
on a strongly optimistic note that the basic error mechanisms
have been identied, and that accurate, replicable particle
data are within reach (67). These continuing advancements
and problems will be reviewed in this section.
Two major advantages of electronic particle counters are
their automated characteristics and the rapidity at which they
accomplish particulate measurement. Two major disadvan-
tages hinder electronic particle analysis frombecoming a more
acceptable means of measuring particulate contamination:
They cannot differentiate among various types of particles,
and they measure particle size differently from microscopic
methods. These advantages and disadvantages are described
in greater detail after rst looking at the basic principles of
different types of electronic particle-counting methods.
Principle of Electrical Resistivity (Coulter Counter)
The Coulter Counter (6870) (Fig. 3.7) detects particles by
measuring the change in electrical resistance produced when
240 Chapter 3
Fig. 3.7 Photograph of Coulter Counter model ZM. (Courtesy
Coulter Electronics, Inc., Hialeh, Florida.)
a particle displaces a part of the electrolyte solution residing
between two electrodes. The change in resistance is directly
proportional to the volume of the particle (70). The Coulter
Counter therefore treats a particle as a three-dimensional ob-
ject. This can be contrasted to the light-blockage principle,
which views a particle in two dimensions and thus calculates
area rather than volume of a particle.
Coulter Counters employ an aperture tube with a known
micrometer opening that is immersed in a volume of paren-
teral solution. The ratio of particle diameter to orice diameter
should be less than 0.3 for the direct proportionality of resis-
tance change and particle volume to be valid (71). For exam-
ple, a 50-mor greater aperture tube should be used for count-
ing particles in the 10-m size range.
Particle analysis must take place in a controlled clean
roomunder high-efciency particulate air (HEPA)ltered air
to minimize environmental particles entering the sample solu-
tion. The aperture tube is immersed in the intravenous solu-
tion, which must be electrolytic. If not, an electrolyte solution
(e.g., sterile sodium chloride for injection) must be added. This
presents a major disadvantage in the use of the Coulter
Counter if an electrolytic solution must be added. The solution
itself may add a signicant number of particles to the sample
solution. Appropriate blank controls must be utilized to sub-
tract the particulate contribution caused by the added electro-
lytic solution.
The instrument employs a manometer to sample from
500 l to .2 ml of the intravenous solution. Counts measured
in this extremely small volume may be too low (1020 parti-
cles) for useful statistical accuracy of sampling (72). Air bub-
bles adversely affect accurate counting. Air bubbles are
avoided by either minimized agitation during sampling or ap-
plication of a vacuum before measurement. Electrical back-
ground noise also contributes to some error in actual counting
of submicronic particles.
Sample solution is pulled through the aperture of the
Coulter Counter solution tube and ows between two elec-
trodes. The change in resistance, proportional to particle vol-
ume, creates a signal that is relayed to a threshold analyzer.
The threshold analyzer has been previously calibrated so that
only pulses of voltage exceeding the threshold position are
counted. The pulses generated are displayed on an oscilloscope
by electronic amplication. Voltage pulse heights are propor-
tional to the amplier gain and aperture current of the instru-
ment and the resistance changes due to the passage of the
particles (24).
The signal produced is proportional to the volume of elec-
trolyte solution displayed by the particle. The count display is
a function of volume directly or the diameter of a sphere of
equal volume (73). The Coulter Counter can count up to 5000
particles per second using the Coulter principle of one-by-one
242 Chapter 3
counting and sizing. Size distributions can be accurarely de-
termined over a range of 0.5 to 800 m, depending on the
proper selection of optimal glassware.
Particulate matter in the subvisible size range present in
intravenous solutions can be detected easily and rapidly by
the Coulter Counter (71,7478). Because of its electrical re-
sistivity principle, the Coulter Counter especially applies in
the determination of particulate contamination in parenteral
electrolyte solutions such as those containing sodiumchloride.
Coulter Counters obtain particle size data with no indication
regarding the shape or composition of the particles. The diam-
eter of particles measured by the Coulter Counter is a mean
spherical diameter. Since particles found in intravenous solu-
tions are usually not spherical, it is important for the orice
dimension of the Coulter aperture tube to be much greater
than the size of particles monitored by the counter. Acicular
particles having lengths much smaller than the diameter of
the aperture orice will produce more accurate pulse heights
having magnitudes closely corresponding to the total volume
of the particle.
Principle of Light Obscuration (HIAC)
A schematic representation of the light obscuration principle
is shown in Fig. 3.8. Atungsten lamp produces a constant colli-
mated beam of light that passes through a small rectangular
passageway and impinges onto a photodiode. In a clear pas-
sageway, the light intensity received by the photodiode re-
mains constant. Liquids can ow through the passageway be-
tween the light source and the photodiode. If a single particle
transverses the light beam, there results a reduction in the
normal amount of light received by the photodiode. This reduc-
tion of light and the measurable decrease in the output from
the photodiode is proportional to the area of the particle inter-
rupting the light ow. Thus, the light-obscuration principle
measures particle size based on the diameter of a circle having
an equivalent area.
Fig. 3.8 Schematic representation of the light obstruction princi-
ple. (Courtesy HIAC/Royco, Menlo Park, California.)
HIAC particle counters employ the light-blockage princi-
ple in the detection and quantitation of particulate matter in
parenteral solutions (see Fig. 3.9). These instruments count
approximately 4000 particles per second. HIAC counters used
sensors having size measurement ratios of 1: 60. In other
words, 1-through-60 micrometer sensor can measure particles
from 1 to 60 m, while a 2.5-through-150 micrometer sensor
can measure particles ranging from 2.5 to 1.50 m. Channel
numbers on the counter are selected and calibrated according
to the size range desired.
Increasingly, over the past several years, HIAC systems
have progressed in technological advances and user applica-
tion in the particle analysis eld. Advances for using HIAC
particle counters have outweighed the disadvantages. Lantz
et al. (79) were among the rst to publish results of HIACanal-
yses of parenteral solution particulate contamination. In con-
clusion, the advantages and disadvantages of using the HIAC
particle counter are given below.
244 Chapter 3
Fig. 3.9 HIAC/Roycos System 8003 for parenteral particle count-
ing to comply with the requirements of microscopic particles. (Cour-
tesy HIAC/Royco Division of Pacic Scientic Company, Silver
Springs, Maryland.)
Advantages
1. Particles are counted automatically.
2. Parenteral solutions, either electrolytes or nonelectro-
lytes, could be counted.
3. The instrument was easy to calibrate and use.
4. Replication of counts was good.
5. The volume of samples could be varied as desired for
counting.
6. Dilution method of counting permitted counting of both
clean and heavily contaminated solution.
7. Direct method of counting permitted counting of crystal-
lized soluble particles.
Disadvantages
1. Instrument is relatively expensive compared to equip-
ment used for counting by optical microscope.
2. Particulate contaminants cannot be identied.
3. Large and/or brous particles may block the sensor
opening.
4. Air bubbles are counted as particulate matter.
5. Dilution method of counting does not permit counting of
crystallized soluble materials because dilution solubilizes
crystals.
Principle of Light Scattering
When a beam of light strikes a solid object, three events occur:
Some of the light is absorbed, some of the light is transmitted,
and the rest of the light is scattered. Scattered light is a com-
posite of diffracted, refracted, and reected light. Particle
counters that operate on the basis of light scattering are de-
signed to measure the intensity of light scattered at xed
angles to the direction of the light beam. A schematic example
is given in Fig. 3.10.
Fig. 3.10 Schematic representation of the light-scattering princi-
ple. (Courtesy Climet Instruments Co., Redlands, California.)
2
4
6
C
h
a
p
t
e
r
3
Table 3.8 Commercially Available Particulate Measurement Systems for Parenteral Use
Model evaluated Mechanism of
Type of system by the FDA
a
measurement Comments
Climet Instruments, Model CI-1000 Light obscuration Excellent large-particle
P.O. Box 1760, Red- detection
lands, CA 92372
Coulter Electronics, Model ZM/P Resistance modulation Not recommended by
13960 NW 60 Street, the FDA for paren-
Miami Lakes, FL terals
33014 Large errors in measur-
ing akes and bers
HIAC/ROYCO, Pacic Model 4103 Light obscuration Past recommendations
Scientic, 2431 Lin- in USP 788
den Lane, Silver
Springs, MD 20910
P
a
r
t
i
c
u
l
a
t
e
M
a
t
t
e
r
T
e
s
t
i
n
g
2
4
7
Kratel Instruments, Boblinger Strasse 23 Light obscuration Good large-particle de-
D7250, Leonburg, tection
Stuttgart, Germany
Met-One, 481 Califor- Model 214 Forward light scatter- Laser diode light
nia Avenue, Grants ing, laser based source
Pass, OR 97526
Particle Measuring Sys- IMOLV/SOPS 100 Forward light scat- Laser diode light
tems, 1855 S. 57th tering source
Court, Boulder, CO
80301
Russel Laboratories, RLV 1-50H Detector only FDA recommended de-
3314 Rubio Crest tector with HIAC
Dr., Altadena, CA
91001
Source: Ref. 81.
a
G. S. Oxborrow, A comparison of particle counters from ve manufacturers, presented at the May 1987 Meeting
on Liquid Borne Particle Inspection and Metrology, Washington, DC.
248 Chapter 3
As a liquid ows into a light-sensing zone, particles in
the uid scatter light in all directions. The scattered light is
directed onto a system of elliptical mirrors that then focus the
light onto a photodetector. The light trap seen in Fig. 3.10 is
designed to absorb most of the main light beam photons.
Met One and Climet particle counters represent exam-
ples of counters operating under this principle. Met One parti-
cle counters are laser-based particle counters that have be-
come very popular instruments in the pharmaceutical
industry both for airborne and liquid-borne particles. For liq-
uid samples, particles in the liquid deect bursts of laser light
energy to a solid-state photodiode in which each burst of light
is converted to a pulse of electrical energy. The electrical
pulses are proportional in height to the particle size. Advan-
tages of laser light particle counts include sample rates of 100
ml per minute, 3500 counts per milliliter, and simultaneous
measurement of six particle sizes.
Davies and Smart (80) reported on rapid assessment of
particle levels in small-volume ampule products with good re-
producibility using the scattered-light-based particle counter.
Advances and disadvantages of counters based on light scat-
tering are similar to those identied for the HIAC counter de-
scribed above. Table 3.8 provides a summary of commercially
available electronic particulate measurement systems.
LIGHT OBSCURATION PARTICLE COUNT TEST
All SVI products containing 100 ml or less are required to pass
the USP test. This includes reconstituted solutions. Products
exempted from the USP test are injections intended solely for
intramuscular and subcutaneous administration or products
for which the label species that the product is to be used with
a nal lter.
Prior to using the USP procedure, three preliminary tests
are to be done:
1. Determination of sensor resolutionUse monosize 10-m
particle standards to ensure that particle size distribution
(manual method) or voltage output distribution (electronic
method) does not exceed the standard particle size by
more than 10%.
2. Sensor ow rateCertify that the actual ow rate is
within the manufacturers specication for the particular
sensor used.
3. Sample volume accuracySince particle count varies di-
rectly with the volume of uid sampled, sampling accu-
racy must be known and be within 5%.
The USP procedure provides specic directions and require-
ments for sample preparation, environmental conditions for
conducting the testing, various necessary equipment to be
used, glassware and closure cleaning, particulate control test,
calibration of the particle-counting instrument, and determi-
nation of particulate matter in the product.
Test Environment
At a minimum, the light obscuration apparatus must be lo-
cated within the connes of a HEPA-ltered laminar ow
workbench. However, most industrial particle test labora-
tories have separate and contained work areas where these
instruments are located. Furthermore, strict standard operat-
ing procedures are followedfor entering the workarea, withper-
sonnel appropriately gowned and samples introduced follow-
ing procedures to remove all extraneous particulate matter.
All glassware, rubber closures, and other samples enter-
ing the testing environment must be scrupulously cleaned as
described in the USP. Final rinsing of all materials is per-
formed using ltered distilled water with the lter porosity
1.2 m or ner.
Prior to determining particle counts of the actual sam-
ples, blank counts are determined using a clean vessel of the
type and volume of the sample. Fifty ml of ltered distilled
water are placed in the vessel; the vessel is agitated, degassed
by sonication, allowed to stand, swirled by hand, and then
three consecutive samples of at least 5 ml are withdrawn and
counted. The rst sample count is discarded, but the second
two sample counts are used to determine acceptability of the
250 Chapter 3
environment. USP requires that there are no more than 10
particles of 10 m or larger and no more than 2 particles 25
m or larger in the combined 10-ml sample.
Test Procedure
Prior to placing test samples within the connes of a laminar
ow work area containing the light obscuration particle
counter, the exterior surfaces of the samples are rinsed with
ltered distilled or deionized water. Since samples need to be
protected from environmental contamination after cleaning
until analyzed, having a specially built clean room or module
offers signicant advantages. Technique to withdraw samples
is very important to minimize particle contamination and in-
troduction of air bubbles. Rubber closed vials may be sampled
either by needles penetrating the closure or by completely re-
moving the closure. This also holds true for vials containing
sterile powders in that diluent can be added either via needle
penetration or rst removing the rubber closure. If samples
are pooled, then it is better to remove the rubber closure.
If the volume of the product in the container is less than
25 ml, then at least 10 units of product are used for the LO
test. Each unit of product is inverted at least 20 times and
shaken vigorously to suspend the particles properly. The 10
units are opened, and the contents are pooled into a cleaned
container. The total volume in the pooled container must be
at least 20 ml. The pooled solution is degassed by sonication
for at least 30 seconds, then left standing so that the air bub-
bles dissipate. The pooled container is then gently swirled and
three aliquots of at least 5 ml of solution are withdrawn and
injected into the LO counter sensor. The rst sample mea-
sured by the electronic counter is not counted, but the next
remaining samples are.
If the volume of the product in the container is 25 ml or
more, the same procedure is followed as above to mix, suspend,
and degas the solutions. Here, only three units of product are
sampled, with each sample at least 5 ml. The rst sample into
the counter is discarded.
Freeze dried and other sterile powdered lled samples
are reconstituted by rst removing the rubber closure without
contamination, using ltered water or other appropriate l-
tered diluent at the required volume. The rubber closure is
then replaced, and the product is manually agitated to ensure
complete dissolution of the drug product. Reconstituted sam-
ples can be pooled at described above, with the total volume of
pooled sample being at least 20 ml. Invert the pooled product
container 20 times prior to withdrawing sample; again, the
rst portion into the counter is discarded.
Calculations
USP provides calculations for pooled and individual samples
from small-volume injections and calculations for individual
unit samples from large-volume injections. The formulas are
Small-volume pooled samples: PV
t
/V
a
n
Small-volume individual samples: PV/V
a
Large-volume individual samples: P/V
where P is the average particle count obtained from the por-
tions analyzed, V
t
is the volume (ml) of pooled sample, V
a
is
the volume of each portion analyzed, V is the volume of the
tested unit, and n is the number of containers pooled.
Interpretation
The injection meets the requirements of the USP subvisible
particle test if the average number of particles present in the
units tested does not exceed the following limits (see Table
3.15 for worldwide compendial requirements)
10 m 25 m
Small-volume injections (particles per container) 6000 600
Large-volume injections (particles per ml) 25 3
If these limits are exceeded, then USP requires that the prod-
uct must be tested by the microscopic particle count test.
252 Chapter 3
MICROSCOPIC PARTICLE COUNT TEST
The present USP method provides both qualitative and quan-
titative data on particulate content in LVI solutions. Particles
not less than 10 m can be counted, sized, and described in
terms of their shape and, at times, their nature (for example,
a cotton ber, piece of glass, or metal sliver). Photographs of
the lter membrane further provide a permanent record of the
particulate test results.
Considerable care and skill are required for preparing the
membrane, cleaning the glassware and equipment used in the
procedure, and using the microscope. This presents a major
disadvantage and motivates pharmaceutical manufacturers
to develop and validate alternative methods employing auto-
mation, electronic counting instrumentation, or both.
Procedure
Laminar Air Flow Hood
All operations and manipulations must be performed under a
certied laminar air ow (LAF) hood equipped with HEPA-
ltered air in a Class 100 environment. Laminar ow hood
certication was discussed in Chapter 1.
Working in a laminar air ow environment can never re-
place the necessity for rigid clean technique in sample prepa-
ration and analysis. Prior to conducting a test, the hood must
be cleaned with an appropriate solvent, preferably 70% etha-
nol or 70% isopropyl alcohol. The HEPA lter itself is not
cleaned because of potential damage to the lter surface.
The hood should have a built-in sink or some accommoda-
tion for collection and disposal of solvent used in the ltration
process.
Introduction and Use of Equipment in the Laminar Air Flow Hood
The USP demands the use of scrupulously clean glassware
and equipment for the particle test. The word scrupulous
means the following:
1. Rinse glassware and equipment successively with (a)
warm detergent solution, (b) hot water, (c) water, and (d)
isopropyl alcohol. The rst supplement of the 19th edition
of the USP listed a fth rinse with trichlorouoroethane
(Freon 113). Freon was eliminated in the 20th edition pro-
cedure because of concern about its toxicity in a closed en-
vironment and harm to the ozone layer.
2. Rinsing technique is important. Glassware and equip-
ment must be rinsed starting at the top of the vertically
held object and working downward in a back-and-forth
manner. Water rinsing may be done outside the LAF hood,
but the nal isopropyl alcohol rinse must be performed
within the hood.
3. After rinsing, all objects must dry under the hood up-
stream of all other operations. This helps to ensure that
few, if any, extraneous particles adhere to the drying ob-
ject.
Rubber Gloves
The USP requires the use of suitable, nonpowdered gloves for
the particle test. Gloves are important in protecting the hands
from the dehydrating effects of isopropyl alcohol. However,
gloves may create more problems than they solve. Using
gloves of improper size will promote problems in careful han-
dling of glassware and equipment. Gloves also produce a false
sense of security, resulting in less than ideally careful manipu-
lations in the LAF hood. The greatest potential limitation of
gloves is the contribution they can make to particulate con-
tamination, even after adequate rinsing. Thus, this require-
ment continues to be controversial.
Membrane Filter and Assembly
Membranes
The USP species the use of a gridded or nongridded, black
or dark gray lter or a lter of suitable material compatible
with the product. The lter must have a porosity of at least
1.0 m.
254 Chapter 3
Explicit instructions are provided in the USP for rinsing
the membrane lter. In the 19th edition of the USP, Freon was
used as the rinsing agent. In the 20th edition, water replaced
Freon. Rinsing of a vertically held lter (using forceps) is ac-
complished using ltered water sprayed from a pressurized
container. Rinsing of the membrane with ltered water starts
at the top of the nongridded side, sweeping a stream of water
back and forth across the membrane surface from top to bot-
tom. This process is repeated on the gridded side of the mem-
brane. Pressure exceeding 2 psi may damage the delicate
membrane.
The rinsing solvent is checked for particle counts, serving
as the blank determination in the testing portion of the USP
procedure. It must be assumed that no dispensing vessel will
provide a particle-free solvent. While the membrane lter on
the nozzle will effectively remove particles above the rated po-
rosity of the lter (usually 1.2 m), particles on the down-
stream side of the lter on the nozzle will shed into the dis-
pensed solvent. Of course, there is always the possibility of a
misplaced or torn membrane lter on the dispenser nozzle.
Filter Assembly
The appropriately rinsed membrane lter is placed with the
grid side up on the lter holder base. Great care is taken when
the ltering funnel is situated on the base so that the mem-
brane is not rumpled or torn. Prior to placing this assembly
on the ltering ask, the unit is rinsed thoroughly and care-
fully with ltered water from the pressurized solvent dis-
penser. After allowing time for the rinse uid to drain the l-
ter, the apparatus is then secured on top of the lter ask.
Test Preparation
Containers to be tested for particulate matter must be in-
verted 20 times before the contents are sampled. Agitation has
been shown to affect particle size distribution (82), so the 20-
fold inversion procedure must be consistent. After rinsing the
outer surface of the container with ltered water, the closure
is removed. One can never be certain that removal of the clo-
sure will not introduce extraneous particles. Careful aseptic
and clean technique must be adhered to as much as possible.
After the closure has been carefully removed, the con-
tents are swirled before 25 ml are transferred to the ltering
funnel. After standing for 1 minute, a vacuum is applied to
lter the 25-ml sample. An additional 25-ml sample of water
is then applied to the sides of the funnel to rinse the walls of
the funnel. The stream of ltered water should not hit the l-
ter membrane for fear of tearing the membrane. The rinse
uid then is ltered via vacuum. Unfortunately, particles tend
to adhere to the underside of the lter assembly top and to
the O rings used between the lter base and lter funnel.
The funnel section of the assembly is carefully removed.
The membrane is lifted away from the base using forceps and
placed on a plastic Petri slide containing a small amount of
stopcock grease or double-sided adhesive tape. The cover of
the Petri slide is placed slightly ajar atop the slide to facilitate
the membrane drying process. The slide then is placed on the
micrometer stage of the microscope for visual analysis.
Particle Count Determination
Examination of the entire membrane lter surface for particu-
lates may be accomplished using a precisely aligned and cali-
brated microscope. The microscope should be binocular, tted
with a 10 objective, and have one ocular equipped with a
micrometer able to measure accurately particles of 10m and
25 m linear dimension. Incident light should be set at an
angle of 10 to 20, although an angle of 30 and 35 has been
reported to be more effective in illuminating the membrane
surface inside a plastic Petri slide (83). Calibration of micro-
scope micrometers based on a National Bureau of Standards
primary standard-stage micrometer has been described by
Lanier et al. (84).
Particles are counted under 100 magnication with the
incident light at an angle of 10 to 20. Obviously, this is a
256 Chapter 3
slow and tedious process requiring patience and dedication on
the part of the microscopist. Use of higher magnication, up
to 400, may be necessary occasionally to discern discrete par-
ticles from agglomerates or amorphous masses (83). Some-
times, particles not visible with dark eld reected light are
very easily observed by means of bright eld illuminization at
45 polarization.
Two sizes of particles are counted, those having effective
linear dimensions 10 m or more and 25 m or more. The
counts obtained from the sample membranes are compared to
counts obtained from a membrane treated exactly like the
sample membrane minus the ltration of the product sample.
Blank membrane counts rarely are zero. However, if 5 or more
particles 25 m or larger and/or more than 20 particles 10
m or larger are counted on the blank membrane, the test is
invalidated, and it signies a serious problem in one or more
of the following areas: poor technique, lter breakdown in the
solvent dispenser, poorly cleaned membranes, poorly cleaned
lter assemblies, and/or HEPA lter leaks. The problem must
be resolved before particle testing can resume. If the USPlimit
of not more than 12 particles per milliliter 10 m or larger
and not more than 2 particles per milliliter 25 m or larger
is exceeded, the large-volume injection product fails the USP
test for particulate matter. For small-volume parenterals, the
test fails if more than 3000 particles/container are 10 m or
larger and/or 300 particles per container are 25 m or larger.
Analysis by microscopic techniques suffers from several
disadvantages: it is very time-consuming, requires technical
expertise, and, because of the manpower requirements, can be
very expensive, The major method for determining subvisible
particulate matter in parenteral solutions, including reconsti-
tuted sterile powders, is the light obscuration technique. How-
ever, if any dispute arises regarding fulllment of USP partic-
ulate matter specications, such disputes must be settled by
applying the ofcial USP microscopic method.
LVI particulate matter standards in other countries gov-
erned by other compendia are reviewed in a section at the end
of this chapter.
AUTOMATED MICROSCOPIC PARTICULATE MATTER TEST
Human variables unavoidably decrease the accuracy, preci-
sion, and reproducibility of manually measuring particulate
matter using the USP microscopic method. As pointed out by
Clements and Swenson (85), considerable time and concentra-
tion are required even under ideal circumstances for the mi-
croscopist to perform the necessary operations to obtain a
clear and accurate view of a particles longest dimension.
Viewing multiple particles adds to the complexity of time and
concentration. Parallax errors (differences in sizing a particle
when seen fromtwo different points not on a straight line with
the particle) decrease accuracy, especially when measuring
smaller (e.g., 510-m) particles. In summary, the success of
application of the manual microscopic technique is directly
proportional to the microscopists efciency, which in turn is
dependent on his or her speed, concentration, and alertness.
To minimize or eliminate the human factor in the USP
particulate matter test, a number of electronic particle-count-
ing instruments have been developed, rened, and computer-
ized for rapid and relatively accurate and reproducible particle
measurements. However, the main disadvantage of electronic
particle counting from a regulatory point of view is a lack of
adherence to the ofcial USP test. As previously noted, elec-
tronic instrumentation for particle counting is permitted to
satisfy USP particle test requirements for large-volume in-
jectables. However, in cases of controversy, the USP micro-
scopic method must be the nal judge.
The nearest equivalent automated system to the USP
manual microscopic method is a system called image analysis.
The system described here differs from the automated inspec-
tion systems described in the section, Visual Inspection: Au-
tomatic Methods. Image analysis is not a 100% nal con-
258 Chapter 3
tainer inspection system. Rather, this system introduces
automation after the large-volume parenteral sample has
been ltered and the membrane prepared. Particle analysis
of the lter membrane is performed by a computer-controlled
microscope and television system.
The Quantimet automated image analysis system has
been fully explained in the literature (85). According to the
USP procedure, the membrane lter, following ltration, is
mounted in a plastic lter holder and placed securely on the
microscope stage plate. An external ber-optic illuminator
provides low-angle, high-intensity illumination with direc-
tional control to satisfy USP requirements and to create opti-
mumparticle contrast against the lter background. Optically
interfaced to the microscope is a high-resolution television
camera. Each eld of the lter surface imaged by the micro-
scope is scanned by the camera, which produces a digital pic-
ture containing geometric and densitometric information. The
camera signal is processed by the central processor, in which
data representing the longest dimension of each particle on
the lter surface are fed to the output computer for processing
and presentation. Results can be displayed on the video moni-
tor printed to provide a permanent hardcopy record, and
stored on magnetic disks for future recall. To meet USP re-
quirements, data generally are reported as the number of par-
ticles having effective linear dimensions equal to or greater
than 10 m and equal to or greater than 25 m.
Millipore Corporation developed a particle measurement
computer system similar in theory, instrumentation, and ap-
plication to the Quantimet system (86). The MC system con-
sists of (a) a microscope and television camera that illuminate
and observe the sample on the lter membrane, (b) a computer
module that receives the video signals fromthe television cam-
era and applies the appropriate logic to count and measure
particles in the viewing area, and (c) a viewing monitor that
subsequently receives the video signal, reconstructs the eld
of view, and prints the desired particle data at the top of the
monitoring screen.
The advantages of automated microscopic analytical sys-
tems are
1. They conform to the USP procedure for particle analysis
of large-volume injectable solutions.
2. Particles are counted, sized, and shape-characterized with
much greater speed and precision compared with the man-
ual microscopic method.
3. Efciency and reproducibility are increased, while tedium
is eliminated.
4. Permanent records in the form of particle data and photo-
micrographs can be obtained.
5. Operation of these systems requires minimal technical
and manipulative skills.
Barber et al. have published a number of interesting arti-
cles attempting to improve methodology for conducting partic-
ulate matter evaluation in parenteral solutions. They have
criticized and suggested improvements in the USP particulate
test (87) and have suggested new methods, such as automated
microscopy (88).
Developments in video-camera technology and image-
processing software have made possible the development of
automatic systems for measuring particle size, counting parti-
cles in various size ranges, and even classifying particles ac-
cording to shape or elemental content (if the imaging system
uses an electron microscope with elemental detection (89).
Photon-correlation spectroscopy is a laser-based tech-
nique that detects scattered laser light froma sample and ana-
lyzes individual photon pulses with an autocorrelator. The
random motion of individual particles with respect to one an-
other produces intensity variations because of interference ef-
fects of the laser light. These variations are measured as diffu-
sion coefcients, and from these values, particle sizes can be
calculated if a shape is assumed. Particle counts cannot
be calculated, but particle sizes between 1 nm and 1 m can
be determined without destroying the sample. Also, measure-
ments are absolute; no calibration is necessary.
260 Chapter 3
Halographic imaging is another laser-based technique in
which halographic images of particles (5 m and larger) in so-
lutions can be measured. This technique also permitted three-
dimensional shape-mapping of particles and holds consider-
able promise for nondestructive particle detection.
Sedimentation eld-ow fractionation employs a centrif-
ugal eld for the separation of particles of different sizes. The
operation of this instrument is similar to that for chromatog-
raphy. The result is a high-effciency separation of a particle
mixture according to weight. Depending on the experimental
conditions, eld-ow fractionation can separate particles in
the 0.011 m or 1100 m ranges.
COMPARISON OF MICROSCOPIC AND ELECTRONIC
PARTICLE-COUNTING METHODS
The comparisons discussed in this section include methods ca-
pable of quantitating particulate contamination, that is, mi-
croscopic and electronic methods. Comparisons involving vi-
sual inspection, both manual and automated methods, were
discussed in the section Comparison to Other Particle Inspec-
tion Methods.
Difculties in comparing particle-counting methods re-
sult from differences in the way in which different methods
determine particle size and distribution. For example, the mi-
croscopic method measures size as the longest linear dimen-
sion of the particle. The principle of light blockage, utilized by
the HIAC particle counter, expresses size as the diameter of
a circle of equivalent area as the actual area consumed by the
particle. Particle counting by electrical resistance (Coulter
Counter) treats the particle as a three-dimensional object and
measures the volume consumed by the particle. Thus, the mi-
croscope, HIAC, and Coulter Counter methods size particles
in one, two, and three dimensions, respectively.
An excellent theoretical discussion by Schroeder and De-
Luca (73) showed that it is virtually impossible to correlate
instrumental and microscopic particle counts directly for ir-
Table 3.9 Summary of Sphericity Correction Factors Based
on Longest Linear Dimension
D
O
D
H
D
A
D
V
longest horizontal light electrolyte
Shape dimension projection blockade displacement
Sphere 1.00 1.00 1.00 1.00
Cube (1: 1: 1) 1.00 0.90 0.95 0.88
Equant (3: 2: 1) 1.00 0.88 0.81 0.62
Prolate ellipsoid 1.00 0.87 0.61 0.52
(2: 7: 1)
Flake (4: 4: 1) 1.00 0.90 0.81 0.55
Rod (3: 1 diameter) 1.00 0.81 0.62 0.52
Fiber (rigid, 10: 1) 1.00 0.64 0.36 0.25
Source: Ref. 73.
regularly shaped particles. As seen in Table 3.9, as long as the
particle is a sphere, all methods will size the sphere equally.
However, as the particle shape deviates from sphericity, the
size measurement by the three alternate approaches will dif-
fer, sometimes drastically, fromthe value obtained by the USP
microscopic method. For example, if the solution sample con-
tained 50 ellipsoid particles with their longest linear dimen-
sion equaling 10 m, the HIAC will yield a count of 50 0.61
30.5 particles. In fact, this HIAC value may be an overesti-
mate because the 0.61 correction factor considers only size (10
m), not the actual number of particles. Assuming the size-
count relationship follows the conventional log-log relation-
ship, the theoretical HIAC count of 50 ellipsoid 10-m parti-
cles would be only 14.4 particles. Figure 3.10 provides the
explanation. The USP microscopic method follows a log-log
distribution, yielding a straight-line slope between 10 m and
25 m for its pass/fail criteria of 50 particles/milliliter at 10
m and 5 particles/milliliter at 25 m. Assuming the HIAC
method follows the same log-log distribution between 10 m
and 25 m, its slope will be parallel to the USP slope. However,
the HIAC correlation factor for ellipsoid particles theoretically
262 Chapter 3
is 0.61 that of the USP method. Thus, the starting point for
the HIAC method is not 10 m, but 6.1 m at the 50-count
position of the log-log graph. Therefore, following a parallel
relationship with the slope of the USP method, the HIAC
method yields a theoretical particle count value of 14.4 parti-
cles at the point intersecting the vertical line from the particle
size of 10 m.
This same logic can be assumed for the Coulter Counter
method. From Table 3.10 and Fig. 3.11, the size correction fac-
tor of the Coulter Counter for ellipsoid particles is 0.52.
Applying the same log-log relationship for a 50 particle/milli-
liter sample, the Coulter Counter will yield a count of about
9.7 particles 10 m or larger.
Hopkins and Young (90) were the rst investigators to
publish actual particle size and number data from typical par-
enteral solutions analyzed by the microscope, HIAC, and
Coulter Counter methods. Some of their results are repro-
duced in Tables 3.11 and 3.12. Table 3.11 shows that the
Coulter Counter yielded particulate counts that deviated be-
tween 9.2% and 40.8% from counts obtained with the mi-
Table 3.10 Count Comparison Between USP and Electrolyte
Displacement Methods
Displacement
count at
Shape USP count SCF 10 m 6.6 m
Sphere 5 1.00 5.00 14.20
Cube 5 0.88 3.63 10.30
Equant 15 0.62 4.51 12.82
Ellipsoid 10 0.52 1.93 5.49
Flake 5 0.55 1.11 3.16
Rod 5 0.52 0.97 2.75
Fiber 5 0.25 0.15 0.44
Total/milliliter 50 17.30 49.16
Source: Ref. 73.
Fig. 3.11 Log count versus log size corrections for sizing and
counting prolate ellipsoids. (From Ref. 73.)
croscope. The HIAC counts were between 19.2% and 3.8%
from those of the microscope. Table 3.12 demonstrates that
(a) the agreement among the three methods was acceptable,
especially considering their different mechanisms of particle
sizing and the fact that these data all fell well within the Aus-
tralian particle standard (100 particles/milliliter 5 m)
Table 3.11 Comparison of Microscope, Coulter, and HIAC Data,
Showing Total Counts of Particles Larger than 5 m
Sample Microscope Coulter HIAC
5606A, Supplier A 6285 6867 5080
5606A, Supplier B 19,364 23,534 18,000
5606A, Supplier B 5113 7200 5020
5606A, Supplier A 4285 4635 3660
5606A, Supplier A 6675 8715 6930
Source: Ref. 90. Data from Technical Documentary Report ML-TDR-64-72,
Air Force Materials Laboratory, Wright-Patterson Air Force Base, Ohio.
264 Chapter 3
Table 3.12 Particle per Milliliter of Isotonic Saline Solutions
in the Range of 550 m
HIAC Coulter Microscope
MFG A S-1
a
Average 10.8 17.4 11.9
MFG B S-2
a
Average 19.2 15.1 11.4
MFG B S-3 Average 7.4 9.7 8.6
MFG B S-4 9.9 6.6 7.9
10.3 9.2 8.1
8.7 7.4 8.3
9.6 7.7 8.1
Filled water blank Average 0.9
Source: Ref. 90.
a
Air bubbles in solution.
present at that time (USP standard was not ofcial at that
time), and (b) great error is produced in both Coulter and
HIAC assays when no attempt has been made to exclude air
bubbles from the sample solutions. These instruments do
count air bubbles as particles; thus, vacuum techniques must
be applied to eliminate air bubbles before any instrumental
particle counting is performed.
A similar conclusion was reached by Rebagay et al. (91)
in that an automatic particle counter can be used in place of
either a polarizing microscope or an image-analyzing system
for routine particulate matter monitoring of various particle
systems (AC Test Dust, polystyrene spheres, antibiotic, elec-
trolyte and large-volume parenteral solutions). However, to do
this, the particle counter must be carefully calibrated with
particles that possess morphological and optical characteris-
tics similar to the particles of interest. An example of their
data for the measurement of particle content of various intra-
venous infusion solutions is given in Table 3.13.
Lim et al. (92) ltered various small-volume parenteral
solutions and counted particles using the manual counting
method under the microscope and the electronic Millipore
method. In products with relatively few particles, both meth-
P
a
r
t
i
c
u
l
a
t
e
M
a
t
t
e
r
T
e
s
t
i
n
g
2
6
5
Table 3.13 Particulate Matter Determination of Some Intravenous Solutions by Automatic
and Microscopic Methods
Microscopic method
a
Automatic
c
Image
Infusion solutions I II analyzer
b
I II
5% Dextrose 2869 336 2604 180 2936 275 2673 192 1748 172
5% Dextrose 0.2% 2003 127 1928 222 2058 159 1813 125 1223 80
NaCl
5% Dextrose 0.45% 1863 67 1708 119 1642 102 1680 89 879 23
NaCl
Lactated Ringers solu- 2078 304 2009 200 2096 190 2039 156 1032 105
tion
0.9% Sodium chloride 1247 136 1201 99 1205 271 1250 201 705 176
10% Protein hydrolysate 7374 267 7408 231 4509 160 7185 879 4252 507
Source: Ref. 91.
a
Reichert Zetopan Universal Microscope. I, Incident polarized light and polycarbonate as substrate; II, incident
bright-eld lighting and cleared white cellulosic substrate.
b
TIMC computer measurement method with cleared white cellulosic substrates.
c
HIAC counter, calibrated with (I) AC Fine Test Dust and (II) polystyrene-divinylbenzene spheres.
266 Chapter 3
ods gave similar results. In products containing a high num-
ber of particles in the size range of 525 m, the electronic
method detected more particles. These authors concluded that
the electronic method was preferable because of its greater
rapidity and precision. A somewhat similar conclusion was
made by Blanchard et al. (93) in comparing the microscope
and the Prototron laser beam (using the light-scattering prin-
ciple). With solutions containing abundant particles of the
small-size range, the particle counter gave more reliable and
accurate results than did the microscope.
CURRENT ISSUES WITH ELECTRONIC
PARTICLE COUNTERS
Knapp and DeLuca (81) listed problems encountered to some
degree with all available instrumentation used to measure
particulate matter (Table 3.14). Many of these problems were
also discussed by Knapp (94), who also proposed action steps
to overcome these problems. Today, many of these problems
have been resolved with advances in the instrumentation
available.
Table 3.14 Problems Encountered with Electronic Particle
Counting Systems
Low sample volume handling capacityimposes sampling errors
due to loss of particles during sampling manipulation before
and during analysis
Shape-dependent signals
Inadequate particle size range
Specication of inappropriate measurement limitsnumber of
particles counted too low for measurement accuracy or the
concentration of particles is too high for sensor capability
Inability to distinguish between particles, microbubbles, and
insoluble microdroplets (e.g., silicone oil)
Flow problems in sensing zone, resulting in random orientation of
particles
Source: Ref. 81.
Barber (95) published an excellent paper detailing the
limitations of LO particle counters as required by the USP for
measuring particulate contamination in SVIs. The single
greatest obstacle in using LOcounters is their inaccurate mea-
surement of both particle number and particle size. This is
not because of design aws or engineering defects with these
counters, but rather because of the basic principle on which
these instruments operate. As discussed in the section Princi-
ple of Light Obscuration, particle counts result from a series
of interactions between a particle moving at high velocity and
an intense light beamin the counters sensor. Whenever a par-
ticle crosses the light beam, the intensity of light that reaches
the photodiode is reduced, and an amplied voltage pulse is
produced. The amplitude of the pulse is approximately propor-
tional to the area of the particle projected onto a plane normal
to the light beam, and the particle size is recorded by the
counter as the diameter of a sphere having an equivalent pro-
jected area. When particles are few, large (5 m), and spheri-
cal, good numerical accuracy is possible. However, when parti-
cles are many, small (5 m), and nonspherical, inaccuracies
will result. A particles residence time in the view-volume usu-
ally is too short to allow the sensor to detect more than one
aspect of the particle; consequently, the LO measurement is
based on the light that is obscured by the particle according
to its orientation when it enters the counters view-volume.
Solution ow rates greatly affect count accuracy. Slower
rates result in longer pulse durations, increased probability of
electronic noise effects on count pulse and possible increases
in apparent particle size. Faster ow rate pulses may not rise
to full height, resulting in undersizing (95).
Nonspherically shaped particles produce signicant er-
rors in the sizing accuracy of electronic particle counters. Be-
cause particles of irregular shape are viewed in randomaspect
as they pass through the sensor of a counter, the size recorded
typically will be less than that dened by the maximum area
of light obscuration. Such an effect is shown in Table 3.9.
As differences between the refractive index of the particle
268 Chapter 3
and the refractive index of the solution containing the particle
increase, the measured particle size will increase. Aparticle in
water will have a greater refractive index, between the two
than the same particle in a concentrated solution of dextrose.
Thus, these particles in water will be measured by the light
obscuration sensor to have greater size and greater number
than the same particles in the concentrated dextrose solution.
Calibration errors can occur because calibration is done
with monosize spherical latex particles, which provide a very
narrow range of known monoshape particle size. This intro-
duces a calibration bias when measuring actual and largely
unknown sizes and shapes of particles in parenteral solutions.
The error introduced nearly always results in particle mea-
surements being smaller than they should be. However, to at-
tempt to calibrate counters with nonspherical particles adds
greater difculties because of their nonuniformity, dispersal
difculties, and differences in chemical composition and opti-
cal properties; the calibration value would be practically
meaningless.
Coincidence effects occur when two or more particles are
counted as a single larger particle. This problem can be most
easily detected by comparing dilutions of the same sample; if
an increase in total counts occurs with the diluted sample, co-
incidence counts are probably the cause. Eradication of coinci-
dence effects is difcult; the only reasonable method for ob-
taining valid data with such solutions is to do microscopic
analyses.
Immiscible uids and air bubbles counted as particles are
other sources of error for light obscuration and other electronic
particle-counting methods. The primary source of immiscible
uid is silicone, usually very small ( 1 m) microdroplets.
Only in signicant numbers do silicon microdroplets produce
signicant errors in particle measurement. Air bubbles are
also problematic, but the USP provides for a method of de-
gassing the sample using ultrasonication. Such degassing
does not remove all microscopic bubbles or reduce the dis-
solved air content in the solution.
Sampling variability, as with any quality control test re-
lying on sampling procedures, must also be recognized as a
source of error with electronic particle counting. Sampling-
associated factors that adversely affect particle counting are
caused by particle stratication effects, by a small sample vol-
ume relative to the total sample volume, and by the low num-
bers of particles per milliliter that typically are counted in a
parenteral solution. Adequate agitation of the product con-
tainer prior to collecting samples must be properly done to
minimize the effects of sampling variability.
In August 1990, the USP invited the Particulate Matter
Committee of the PMAQCSection to meet and discuss various
suggestions for improving the USP test for particulate matter
(Section 788) (96). The USP had a three-stage proposal: (a) an
increase in the minimum volume of sample to be tested, (b)
an increase in the minimum number of test units to be tested,
and (c) use of composite versus individual testing. The com-
mittee decided that none of these changes needed to be made.
However, the committee did agree to do three new assign-
ments: (a) provide data and information regarding the possi-
bility of tightening the light obscuration limits in the USP, (b)
seek and provide silicone assays that could be used for testing
at Stage 2, and (c) study the adoption and the Improved Micro-
scopic Assay for Stage 3.
FACTORS AFFECTING ACCURATE PARTICLE TESTING
Nearly every scientic paper featuring the use of a particle
test method, be it visual, microscopic, electronic, manual, or
automatic, has to alert the reader to one or more major limita-
tions to the method. These limitations have been addressed
in this chapter. For example, visual examination by human
beings is limited by its tedium and subjectiveness. Micro-
scopes often are improperly calibrated. Electronic particle
counters count air bubbles as particles. For LVIs, the USP re-
lies on membrane ltration, by which particles fromthe equip-
ment, environment, or personnel involved in conducting the
test inadvertently become deposited on the lter.
Other problems exist that can potentially cause inaccu-
270 Chapter 3
rate particle test results regardless of the test used. Ernerot
(97) pointed out that the particle contents of injection contain-
ers vary considerably between the date of production and a
later date when the same containers are tested again. It was
found that storage causes particle agglomeration. Mechanical
agitation breaks up the agglomerates, resulting in counts that
cannot reproduce the original count or replicate one another
on the same date of testing. Freshly prepared solutions
seemed to give more stable counts. It was suggested that
only the manufacturer, who can reproduce the handling of
its products, use particle counting as a meaningful control
method.
Agitation or shaking will increase the number of particles
in a parenteral solution. Blanchard et al. (82) found that the
slope and number of particles per milliliter greater than 1 m
in a log-log plot of number against diameter depended on the
degree of agitation. Agitation of LVI by 20 hand inversions,
as required by the USP procedure, removed particulate matter
from the surface of the container, thus increasing the total
number of particles greater than 1 m. Yet, the relative size
distribution of particles was not altered signicantly. Agita-
tion for 30 minutes disintegrated agglomerates, greatly in-
creased the number of particles with diameters less than 1 m,
and brought about a corresponding decrease in the number of
particles exceeding 1 m in diameter. Particle-counting pro-
cedures must be carried out that do not impose a sheer force
on the particles and affect the reproducibility of the test re-
sults.
Temperature affects the number of particles found in par-
enteral solutions. As shown in Fig. 3.12 (98), particle number
increased as a function of temperature and time. Interestingly
and without clear explanation, a decrease in the particle num-
ber occurred after 120 hours of storage at all temperatures.
Since the particle size range studied was 2.33 to 5.02 m (us-
ing the Coulter Counter), it is possible that particle agglomer-
ation occurred, resulting in a decrease in particle number at
these smaller diameters, but an increase in particle counts at
Fig. 3.12 Number of particles between 2.33 and 5.02 m found in
resting plastic bags after shaking for 30 hours as a function of time
and temperature. Key: , 35C; , 45C; , 55C; , room tempera-
ture. (From Ref. 98.)
larger sizes. These same investigators found that glass con-
tainers produced fewer particles than plastic containers under
similar storage and handling conditions.
INTERNATIONAL COMPENDIA STANDARDS
FOR PARTICULATE MATTER CONTENT
IN PARENTERAL SOLUTIONS
Table 3.15 is based on an article by Lotteau (99) and a review
of the available compendia updated to reect requirements as
of 2002. In comparing the requirements from different com-
pendia, the following should be noted:
1. The EP now has particulate matter requirements for
small-volume injectables that were effective in 2000.
2
7
2
C
h
a
p
t
e
r
3
Table 3.15 Comparison of Compendia for Particulate Matter Standards
Compendia LVI/SVI Method Limits*
USP LVI Microscopic 12 parts/ml 10 m
2 parts/ml 25 m
SVI Light obscuration 6000 parts/container 10 m
600 parts/container 25 m
BP LVI Coulter Counter 1000 parts/ml 2 m
European Solutions 100 ml Light obscuration 25 parts/ml 10 m
Test A 3 parts/ml 25 m
Solutions 100 ml Light obscuration 6000 parts/container 10 m
Test B 600 parts/container 25 m
Japan LVI Microscopic 20 parts/container 10 m
2 parts/ml 25 m
SVI Light obscuration 1000 parts/container 10 m (proposed)
* parts particles.
2. The EP differentiates particulate matter requirements in
small-volume injectables whether the injectable product
is a ready-to-use solution or a powder that has been recon-
stituted into a solution.
3. Only the BP designates the use of the Coulter Counter as
the electronic method for measuring particulate matter in
large-volume solutions.
4. Each compendia has slightly different acceptability stan-
dards with respect to either particle size and/or particle
number.
REFERENCES
1. U.S. Pharmacopeia, 25th edition, Mack Publishing, Easton,
Pennsylvania, 2002, Section 1 Injections.
Pennsylvania, 2002, Section 788 Particulate Matter in In-
jections.
3. European Pharmacopeia, Supplement 2002, pp. 219225.
4. T. A. Barber, Control of Particulate Matter Contamination in
Healthcare Manufacturing, Interpharm Press, Denver, CO,
2000.
5. E. W. Godding, Foreign matter in solutions for injection,
Pharm. J., 154, (1945).
Pennsylvania, 1947.
7. J. H. Brewer and J. H. F. Dunning, An in vitro and in vivo
study of glass particles in ampuls, J. Am. Pharm. Assoc., 36,
289 (1947).
8. N. Leelarasamee, S. A. Howard, and H. J. Baldwin, Visible par-
ticle limits in small-volume parenterals, J. Parenter. Drug
Assoc., 34, 167174 (1980).
9. B. E. Knowaler, Pulmonary emboli of cotton bers, Am. J. Clin.
Pathol., 20, 385 (1950).
10. W. B. Wartman, B. Hudson, and R. B. Jennings, Experimental
arterial disease: the role of the pulmonary artery to emboli of
lter paper bers, Circulation, 4, 756 (1951).
274 Chapter 3
11. E. Gnadinger, A new procedure for the control of injection solu-
tions, thesis, University of Strasbourg, 1957.
12. J. M. Garvan and B. W. Gunner, Intravenous uids: a solution
containing such particles must not be used, Med. J. Austral.,
2, 140 (1963).
13. J. M. Garvan and B. W. Gunner, The harmful effects of parti-
cles in intravenous uids. Med. J. Aust., 22, 1 (1964).
14. R. M. Geisler, The biological effects of polystyrene latex parti-
cles administered intravenously to rats: a collaborative study,
15. K. Hozumi, K. Kitamura, T. Kitade, and S. Iwagami, Localiza-
tion of glass particles in animal organs derived from cutting of
glass ampoules before intravenous injections, Microchem. J.,
28, 215226 (1983).
16. J. E. Dimmick, Fiber embolization: a hazard of cardiac surgery
and catheterization, N. Engl. J. Med., 292, 685 (1975).
17. P. P. DeLuca, R. R. Rapp, B. Bivins, H. E. McKean, and W. O.
Grifn, Filtration and infusion phlebitis: a double-blind pro-
spective clinical study, Am. J. Hosp. Pharm., 32, 101 (1975).
18. S. Turco and R. E. King, Sterile Dosage Forms, 3rd ed., Lea and
Febiger, Philadelphia, Pennsylvania, 1987, pp. 372378.
19. H. Stein and B. Vu, Coring: a potential problem in eye surgery,
J. Cataract Refract. Surg., 20, 169171 (1994).
Healthcare Manufacturing, Interpharm Press, Denver, CO,
2000, p. 9.
21. L. J. Pesko, Physiological consequences of injected particles,
Liquid and Surface Borne Particle Measurement Handbook
(J. Z. Knapp, T. A. Barber, and A. W. Lieberman, Eds.), Marcel
Dekker, New York, 1996.
22. R. J. Duma, M. J. Akers, and S. Turco, Parenteral drug admin-
istration: routes, precautions, problems, and complications,
Pharmaceutical Dosage Forms: Parenteral Medications, 2nd
Ed. (K. E. Avis, L. Lachman, and H. Lieberman, Eds.), Marcel
Dekker, New York, 1991.
23. P. P. DeLuca, Particulate matter in parenteral products,
Pharm. Tech., 3, 30 (1979).
24. M. J. Groves, Parenteral Products, William Heinemann, Lon-
don, 1973.
25. R. G. Draftz and J. Graf, Identifying particle contaminants,
26. W. C. McCrone, Identication of parenteral particulate con-
taminants, Pharm. Tech., 2, 57 (1978).
Healthcare Manufacturing, Interpharm Press, Denver, Colo-
rado, 2000, p. 6196.
28. U.S. Pharmacopeia, 8th ed., Blakiston, Pennsylvania, 1905.
29. G. Schwartzman, Particulate matter as viewed by the FDA,
Bull. Parenter. Drug. Assoc., 31, 161 (1977).
30. National Formulary, 6th Ed., Mack Publishing, Easton, Penn-
sylvania, 1936.
31. First Supplement to U.S. Pharmacopeia (19th Ed.) and Na-
tional Formulary (14th Ed.), USP Convention, Inc., Rockville,
Maryland, 1975, pp. 5557.
32. I. Vessey and C. E. Kendall, Determination of particulate mat-
ter in intravenous uids, Analyst, 91, 273 (1966).
33. Proposed General Standard for Injections, Particulate Matter
in Large-Volume Injections, National Biological Standards
Laboratory, Canberra, Australia, 1974.
34. British Pharmacopoeia, Vol. 2, Appendix 13, Her Majestys Sta-
tionary Ofce, London, England, 1980, p. A120.
35. D. M. Hailey, A. R. Lea, and C. E. Kendall, Specications of
limits for particulate contamination in pharmaceutical dosage
forms, J. Pharm. Pharmacol., 34, 615620 (1982).
36. M. J. Groves, Modern techniques for the control of particulate
contamination in parenteral solutions, Cron. Farm., 20, 212
217 (1977).
37. M. J. Groves and D. Wana, A comparison of instrumental tech-
niques for the determination of contamination in ltered aque-
ous solutions, Powder Tech., 18, 215223 (1977).
38. W. E. Hamlin, General guidelines for the visual inspection of
parenteral products in nal containers and in-line inspection of
container components. J. Parenter. Drug. Assoc., 32, 63 (1978).
276 Chapter 3
39. The Pharmacopoeia of Japan, 13th Ed., Takeji Nippo, Ltd.,
Tokyo, p. 21.
40. W. D. Graham, L. G. Chatten, M. Pernarowski, C. E. Cox, and
J. M. Airth, A collaborative study on the detection of particles
in amputed solutions, Drug Standards, 27, 61 (1959).
41. H. J. Baldwin, S. A. Howard, and N. Leelarasamee, Visual
ampul inspection by experienced and non-experienced inspec-
tors. J. Parenter. Sci. Tech., 35, 148 (1981).
42. D. Schipper and R. Gaines, Comparison of optic electronic in-
spection and manual visual inspection, J. Parenter. Drug
Assoc., 32, 118 (1978).
43. A. C. V. M. Faesen, Reproducibility of visual inspection of par-
enterals, J. Parenter. Drug Assoc., 32, 75 (1978).
44. P. R. Rasanen and R. C. Louer, Panel discussion: mechanical
inspection of ampuls Iampul inspection using Autoskan
equipment, Bull. Parenter. Drug Assoc., 27, 243 (1973).
45. A. B. McGinn, Panel discussion: mechanical inspection of
ampuls IIampul inspection using the Rota machine. bull.
Parenter. Drug Assoc., 27, 247 (1973).
46. R. J. Walsh and J. A. Monahan, Panel discussion: mechanical
inspection of ampuls IIIampul inspection using the Strunck
machine, Bull. Parenter. Drug Assoc., 27, 250 (1973).
47. J. Blanchard, C. M. Thompson, and J. A. Schwartz, Compari-
son of methods for detection of particulate matter in large-vol-
ume parenterals, Am. J. Hosp. Pharm., 33, 144 (1976).
48. J. B. Wynn, Mechanization of ampoule inspection, Bull. Pare-
nter. Drug Assoc., 22, 13 (1968).
49. G. W. Martyn, Utilization of Rondo unscrambler and Strunck
units in ampoule inspection, Bull. Parenter. Drug Assoc., 24,
281 (1970).
50. S. Levine, Particulate matter: III. Electronic inspection, Bull.
Parenter. Drug Assoc., 20, 33 (1966).
51. W. E. Hamlin, W. W. Scothorn, and K. W. Riebe, Experience
with a prototype television inspection device: I. Detectability of
subvisible particles, Bull. Parenter. Drug Assoc., 31, 86 (1977).
52. B. H. Kaye, Statistical strategies and ROEPA (robot for exam-
ining parenterals), Bull. Parenter. Drug Assoc., 33, 239 (1979).
53. M. J. Groves, Detection of particles in intravenous solution us-
ing the nucleopore prototron counter, Pharm. J., 213, 518
(1974).
54. M. C. Porter, New particle counter for nondestructive monitor-
ing of parenterals, Pharm. J., 29, 169 (1975).
55. R. C. Louer, J. A. Russoman, and P. R. Rasanen, Detection of
particulate matter in parenteral solutions by image projection:
a concept and its feasibility, Bull. Parenter. Drug Assoc., 25, 54
(1971).
56. J. Z. Knapp and H. K. Kushner, Implementation and automa-
tion of a particle detection system for parenteral products. J.
Parenter. Drug. Assoc., 34, 369 (1980).
57. J. Z. Knapp, The scientic basis for visible particle inspection,
PDA J. Pharm. Sci. Tech., 53, 291301, 1999.
58. J. Z. Knapp and H. K. Kushner, Generalized methodology for
evaluation of parenteral inspection procedures. J. Parenter.
Drug. Assoc., 34, 14 (1980).
59. J. Z. Knapp, H. K. Kushner, and L. R. Abramson, Automated
particulate detection for ampuls with the use of the probabilis-
tic particulate detection model, J. Parenter. Sci., Tech., 35, 21
(1981).
60. J. Z. Knapp, H. K. Kushner, and L. R. Abramson, Particulate
inspection of parenteral products: an assessment, J. Parenter.
Sci. Tech., 35, 176 (1981).
61. J. Z. Knapp and H. K. Kushner, Particulate inspection of paren-
teral products: from biophysics to automation, J. Parenter. Sci.
Tech., 36, 121 (1982).
62. C. H. Rothrock, R. Gaines, and T. Greer, Evaluating different
inspection parameters, J. Parenter. Sci. Tech., 37, 64 (1983).
63. J. Beaman, Pzer Global R&D, personal communication, Janu-
ary 2001.
64. FDA Warning Letters. Available at www.pharmcast.com and
www.fda.gov.
278 Chapter 3
65. Particulate matter standards for small-volume injections,
Pharm. Forum, 37293735 (NovemberDecember 1983).
66. International Conference on Liquid-Borne Particulate Inspec-
tion and Metrology, Washington, D.C., May 1113, 1987.
67. J. Z. Knapp and T. A. Barber, overview of the international
conference on particle detection, metrology and control, J. Pare-
nter. Sci. Tech., 257263 (1990).
68. W. H. Coulter, U.S. Patent 2,656,508 (1953).
69. Coulter Electronics, Inc., Theory of the Coulter Counter, Bulle-
tin T-1, 1958.
70. R. Cartmel and H. N. Gerrard, The Coulter Counter, Pharm.
J., 191, 383 (1963).
71. O. Q. Ulbrich, Size Analysis of Fine Particles and Results Ob-
tained with an Electrical Sensing Zone Particle Analyzer, pa-
per presented before Instrument Society of American Confer-
ence, New York, September 26, 1960.
72. S. M. Blaug and R. E. Sarabia, Determination of particulate
matter in intravenous solutions using the Coulter Counter,
73. H. G. Schroeder and P. O. DeLuca, Theoretical aspects of par-
ticulate matter monitoring by microscopic and instrumental
methods, J. Parenter. Drug Assoc., 34, 183 (1980).
74. M. J. Groves and J. F. G. Major, Assessment of particulate ma-
terial in normal saline solution, for injection B.P. by means of
the Coulter Counter, Pharm. J., 193, (1964).
75. R. W. Lines, An insertable orice tube for in situ contamination
counts of solutions in opened ampoules and vials using the
Coulter Counter, J. Pharm. Pharmacol., 19, 701 (1967).
76. M. J. Groves, Coulter Counter estimation of particles retained
on terminal ltration units, Pharm. J., 208, 550 (1972).
77. T. G. Somerville and M. Gibson, Particulate contamination in
ampuls, Pharm. J., 211, 128 (1973).
78. A. B. Bikhazi, J. A. Shiatis, and A. F. Haddad, Quantitative
estimation of particulate matter in pharmaceutical prepara-
tions intended for intravenous administration, J. Pharm. Sci.,
66, 181 (1977).
79. R. J. Lantz, E. G. Shami, and L. Lachman, Application of the
HIAC particle counter to monitoring particulate matter in par-
enteral solutions I: a preliminary study, Bull. Parenter. Drug
Assoc., 30, 234 (1976).
80. P. J. Davies and J. D. Smart, Evaluation of the Royco 367
small-volume sample feeder for estimation of particulate con-
tamination in ampoules, J. Pharm. Pharmacol., 31, 108P
(1979).
81. J. Z. Knapp and P. P. DeLuca, Review of commercially avail-
able particulate measurement systems, J. Parenter. Sci. Tech.,
Technical Report No. 9, 42 Skupp. (1988).
82. J. Blanchard, J. A. Schwartz, and D. M. Byrne, Effects of agita-
tion on size distribution of particulate matter in large-volume
parenterals, J. Pharm. Sci., 66, 935 (1977).
83. M. A. Wente, Particulate matter levels in large-volume inspec-
tions: USP 19 method, Bull. Parenter. Drug. Assoc., 31, 165
(1977).
84. J. M. Lanier, G. S. Oxborrow, and L. T. Kononen, Calibration
of microscopes for measuring particles found in parenteral so-
lutions, J. Parenter. Drug Assoc., 32, 145 (1978).
85. B. Clements and R. Swenson, Automated microscopy for con-
tamination control, Pharm. Tech., 22, 42 (1978).
86. J. W. Warren, Jr., T. E. Needham, and J. D. Benmaman, Partic-
ulate matter detection in LVIs and I.V. sets: automated micro-
scopic and Coulter Counter methods, PharmTech., 2, 31 (1978).
87. T. A. Barber, M. D. Lannis, J. G. Williams, and J. F. Ryan,
Application of improved standardization methods and instru-
mentation in the USP particulate test for SVI, J. Parenter. Sci.
Tech., 44, 184203 (1990).
88. T. A. Barber, M. D. Lamis, J. G. Williams, Method evaluation:
automated microscopy as a compendial test for particulates in
parenteral solutions, J. Parenteral Sci. Tech., 43, 2747 (1989).
89. J. E. Freeman, Subvisible Particle Detection, Conference on
Visible and Subvisible Particles in Parental Products, Euro-
pean Organization of Quality Control, Brussels, October 30
31, 1985.
90. G. H. Hopkins and R. W. Young, Correlation of microscopic
280 Chapter 3
with instrumental particle counts, Bull. Parenter. Drug Assoc.,
28, 15 (1974).
91. T. Rebagay, H. G. Schroeder, and P. P. DeLuca, Particulate
matter monitoring II: Correlation of microscopic and automatic
counting methods, Bull. Parenter. Drug Assoc., 31, 150 (1977).
92. Y. S. Lim, S. Turco, and N. M. Davis, Particulate matter in
SVIs as determined by two methods, Am. J. Hosp. Pharm., 30,
518 (1973).
93. J. Blanchard, J. A. Schwartz, D. M. Byrne, and D. B. Marx,
Comparison of two methods for obtaining size distribution
characteristics of particulate matter in large-volume parenter-
als, J. Pharm. Sci., 67, 340 (1978).
94. J. Z. Knapp, Particulate matter counting: the technical issues,
J. Parenter. Sci. Tech., 42, 28 (1988).
95. T. A. Barber, Limitations in light-obscuration particle counting
as a compendial test for parenteral solutions, Pharm. Tech. 34
52 (October 1988).
96. J. A. Syverson, Small-Volume Injectables Particulate Matter
Committee, presented at Pharmaceutical Manufacturers Asso-
ciation Quality Control Meeting, Washington, D.C., March 8,
1991.
97. L. Ernerot, Particle counting in infusion uids, Acta Pharm.
Suec., 11, 1 (1974).
98. R. J. Whitlow, T. E. Needham, and L. A. Luzzi, Generation of
particulate matter in large-volume parenteral containers, J.
Pharm. Sci., 63, 1610 (1974).
99. P. Lotteau, Control of particle contamination of solutions for
pharmaceutical use: selection of a suitable method, J. Parenter.
Sci. Tech., 44, 302305 (1990).
4
Package Integrity Testing
INTRODUCTION
The assurance of patient safety requires that all parenteral
product packaging prevent the ingress of microbial contami-
nants. While this is certainly critical, it can be said from a
broader perspective that package integrity is the measure of
a packages ability to keep the product in and to keep potential
contaminants out. The product necessitating containment is
typically thought of as either the formulated liquid or the solid
drug product inside the package. But the product may also
include the gas headspace environment within the package.
Such is the case for products requiring the maintenance of a
rareed gas headspace or a partial pressure environment to
ensure either the products stability or the packages function-
ality. And while parenteral packaging must prevent potential
microbial contamination, it must also prevent the ingress of
unwanted environmental debris, chemicals, or particulates.
Package integrity is a requirement to be met throughout
the products life cycle. The verication of a packages ability
281
282 Chapter 4
to maintain adequate integrity should begin in the develop-
ment stages of a product. Package integrity assurance should
continue after product approval by manufacturing within the
acceptable ranges of established critical specications and
process controls. Manufactured product integrity may be sup-
ported by the additional use of appropriate tests conducted
in conjunction with production operations. Shelf life package
integrity tests of stability samples provide nal evidence of
marketed product quality.
Regulations passed primarily by the U.S. Food and Drug
Administration (FDA) have helped to drive the incorporation
of appropriate package integrity tests as part of the develop-
ment process and ultimate manufacture of parenteral prod-
ucts. A variety of tests is available for use by the pharmaceuti-
cal industry to measure parenteral product package integrity.
While all testing approaches have some value, each also has
drawbacks. Test methods vary in sensitivity and capability.
This is complicated by the fact that products differ in package
integrity requirements. Even different seals within the same
product/package system may have different integrity de-
mands. For instance, a package system that includes a paren-
teral delivery device may consist of a closure to contain the
product and to prevent liquid microbial ingress, as well as an
attachment for product administration only intended to pre-
vent airborne microbial contamination at the time of use.
Finally, the stage of the products life cycle may stipulate
different types of testing to be performed; package valida-
tion studies may require more exhaustive and sensitive test-
ing than routine manufacturing. Therefore, no single test can
be recommended for all parenteral package integrity test-
ing.
Fortunately, research and development of package integ-
rity tests has escalated in recent years. By gleaning a general
understanding of leakage concepts and by utilizing published
integrity test data, it is possible to more logically design, as-
semble, and validate integral parenteral product packaging.
Package Integrity Testing 283
REGULATIONS AND GUIDANCES
U.S. Food and Drug Administration
Prior to the mid-1990s only sterility of the packaged product
was required by the FDA as verication of package integrity.
We have seen a dramatic change in the FDAs requirements
for package integrity verication over the last several years,
as several nal and draft industry guidance documents issued
by the FDA include discussions of package integrity. The rst
of these documents was issued in 1994; Guidance for the Sub-
mission Documentation for Sterilization Process Validation in
Applications for Human and Veterinary Drug Products (1).
This landmark guidance required that the microbial barrier
properties of a parenteral product package be veried and
stated that sterility testing alone is insufcient for this pur-
pose. This guidance included the following requirements re-
lating to package integrity:
1. Validated integrity studies must be performed, and data
must be included in the submission.
2. Experiments must stimulate the stresses anticipated dur-
ing product life, including maximum sterilization cycle(s).
3. Physical, chemical tests are allowed as well as the more
traditional microbial challenge studies.
4. Each sterile package seal barrier should be separately
evaluated and validated.
5. The sensitivity of the method used should be specied and
provided.
6. Microbial integrity should be demonstrated over the shelf
life of the product.
This was followed by a Draft Guidance in 1998: Con-
tainer and Closure Integrity Testing in Lieu of Sterility Test-
ing as a Component of the Stability Protocol for Sterile Prod-
ucts (2). This document solely addressed the need for
incorporation of appropriate, adequately validated container
closure integrity tests, beyond sterility testing, to verify a
284 Chapter 4
packages microbial barrier properties as a part of a sterile
products stability protocol. Both microbial challenge methods
as well as physicochemical test methods were permitted. Sta-
bility studies included those for pending new product appli-
cations, investigative or unlicensed products, and approved/
licensed products. Products included biologicals, human and
veterinary drugs, and medical devices. Tests were recom-
mended annually and at product expiry; sterility was still re-
quired for product release and at product expiry. This draft
was never issued as a nal guidance, and very likely it will
not be as package integrity is addressed in later guidance doc-
uments that focus on specic product development stages and
regulatory document submissions. However, it did much to
clarify the agencys thinking on package integrity.
Again in 1998, the FDA issued a Draft Guidance for In-
dustry: Stability Testing of Drug Substances and Drug Prod-
ucts (3). This draft, still in active status at the time of this
chapters preparation, covers stability studies needed
throughout the pharmaceutical products life, including Phase
1, 2, and 3 development and new drug application (NDA) and
abbreviated new drug application (ANDA) submissions. Ap-
proved product stability protocols and stability testing to sup-
port postapproval changes are included as well. Container in-
tegrity is addressed in Section VII (Specic Stability Topics,
C. Microbiological Control and Quality). Here, it is stated that
appropriately sensitive, adequately validated integrity tests
are to be performed annually and at expiry throughout the
products stability program to demonstrate a packages ability
to prevent microbial ingress.
The FDA issued another Draft Guidance for Industry in
February 1999: INDs for Phase 2 and 3 Studies of Drugs, In-
cluding Specied Therapeutic Biotechnology-Derived Product,
Chemistry Manufacturing, and Controls Content and Format
(4). This Draft Guidance recommends the inclusion of con-
tainer closure microbial integrity tests in Phase 3 stability
studies to demonstrate package integrity during drug product
shelf life. Along with the submission of the test to the FDA,
a rationale for test selection relating to container integrity
should be provided.
A nal FDA Guidance for Industry was issued in May
1999: Container Closure Systems for Packaging Human
Drugs and Biologics (5). This guidance document covers the
requirements that must be met for container closure systems
intended for packaging all human drug and biological prod-
ucts. A demonstration of container closure integrity is one
portion of the information required to justify a packages suit-
ability for its intended use. Parenteral packages must demon-
strate integrity against microbial and environmental con-
tamination. But in addition, packages for some products must
prevent solvent loss, exposure to reactive gases, or absorption
of water vapor. The integrity tests used and the data gener-
ated are to be submitted as part of the chemistry, manufactur-
ing, and controls section of a new drug application.
In conclusion, it is clear that the FDA understands that
package integrity is a critical requirement for all pharmaceu-
tical products. As a matter of consumer safety, industry is be-
ing lead by the FDA to verify the ability of parenteral product
packaging to prevent microbial contamination for products in
later stages of clinical development as well as those approved
for marketing. While microbial challenge methods have
traditionally been used for this purpose, the FDA will per-
mit the utilization of other approaches with the submission of
supportive scientic rationale. In addition, integrity verica-
tion beyond microbial barrier testing is required during de-
velopment for products that require the minimization of sol-
vent loss, exposure to reactive gases, or absorption of water
vapor.
European Union Regulatory Bodies
European regulatory requirements say little to date about con-
tainer closure integrity of parenteral or sterile pharmaceutical
products. Regulations provide for package integrity verica-
tion of parenteral vials to be supported by the performance of
286 Chapter 4
sterility tests as part of the stability program. More specic
information is described in the EU 1998 Rules Governing
Medical Products in the European Union, Pharmaceutical
Legislation (6). These GMP regulations require that the seal-
ing or closure process be validated. Packages sealed by fusion
(e.g., ampules) should be 100% integrity tested. Other pack-
ages should be sampled and checked appropriately. Packages
sealed under vacuum should be checked for the presence of
vacuum.
While not as detailed as the FDA Guidances, it is evident
that the EU Rules also require the verication of parenteral
product package seal integrity. It is important to note that the
EU Rules specically require 100% product testing for fusion-
sealed packages, sampling and testing of all other packages,
and vacuum verication for packages sealed under partial
pressure. The FDA Guidances, on the other hand, do not man-
date the extent of testing within a production lot or call for
the verication of vacuum presence.
PDA Technical Report No. 27
Due to the complexity of pharmaceutical product/package sys-
tems and the confusion that can arise when trying to select
the most appropriate integrity tests, an effort was made by
the Parenteral Drug Association (PDA) to publish a technical
resource to offer clarication. The end result was Technical
Report No. 27, Pharmaceutical Package Integrity, prepared
by a task force of scientists from both industry and academia
(7). This report begins with a summary of package leakage
concepts and critical leak specications. It discusses the need
to consider package integrity for the life of the product, begin-
ning in early product development. Eighteen different integ-
rity tests are described and thoroughly referenced. These are
linked to a decision tree to help the reader in selecting the
most appropriate methods. While neither an ofcial regula-
tory requirement nor guidance document, PDA Technical
Report No. 27 is included in this section as it can serve as
a valuable reference when selecting and developing package
integrity tests.
LEAKAGE
Denition
Leakage occurs when a discontinuity exists in the wall of a
package that can allowthe passage of gas under the action of a
pressure or concentration differential existing across the wall.
Leakage differs from permeation, which is the ow of matter
through the barrier itself.
Permeation is governed by Ficks laws of diffusion (Eqs.
1 and 2), for which permeation rate is a function of the perme-
ants concentration and its solubility in the barrier material
as well as the molecules physical ability to migrate through
the barrier (8,9). For permeation to occur, the molecule must
be adsorbed onto the barrier, then move through the material
by dissociation and migration, and nally exit by desorption
on the other side of the barrier.
Ficks rst lawassumes a barrier of innitely small thick-
ness, that is, a membrane:
J D(C/x)
t
, (1)
where
J amount of diffusion, g/m
2
s
D diffusion constant, m
2
/s
C concentration of diffusant, g/m
3
x barrier thickness, m
t time, s
In the case of a barrier of measurable thickness, the con-
centration gradient of diffusant varies across the thickness,
and is continually changing with time, thus acting to change
the ux. This situation is dened by Ficks second law (8,9),
where
C/t D(
2
C/x
2
) (2)
288 Chapter 4
Fig. 4.1 Permeation ux versus time when the time lag for the
penetration of the diffusant is a function of the barrier thickness
L and the diffusion constant D. (Courtesy of the Parenteral Drug
Association, Inc., Bethesda, MD.) (Source: Ref. 9.)
A graphic representation of permeation ux with respect to
time is given in Fig. 4.1.
Leakage, on the other hand, is a mass ow phenomenon
in which molecules move by convection and diffusion through
a gap in the barrier. Solubility of the diffusant in the barrier
material plays no part in either the convection or the diffu-
sional ux associated with leakage. The driving force for con-
vection of gases as well as liquids is the pressure differential
across the pore or leak. The driving force for diffusional leak-
age is the concentration gradient existing across the leakage
gap. This relationship is described in Eq. 3, where the total
ux for species A is a function of the sum of the convective
uxes for all species and of the diffusion of A, which depends
on the concentration gradient of only species A (10):
J
A
X
A
( j
A
j
B
) (p)(D
A
)(X
A
)(1/x) (3)
where
J
A
total leakage ux of species A, g/m
2
s
X
A
fractional amount of species A, dimensionless
j convective ux of each species, A and B, g/m
2
s
p density of mixture, g/m
3
D
A
diffusion coefcient of species A, m
2
/s
x barrier thickness, m
Diffusional Flux, Convective Flux, or Permeation?
Before trying to apply the theories of leakage ux and perme-
ation to a particular package, it may be helpful to consider a
few practical examples of how these phenomena relate to one
another.
One example of leakage diffusional and convective ux
acting simultaneously is a sealed parenteral vial containing a
nitrogen headspace shipped in a partially pressurized cargo
section of an airplane. On the ground under ambient condi-
tions, oxygen will tend to diffuse into and out of the container
through any leaks present, although more oxygen will diffuse
into the container than out due to the partial pressure differ-
ential existing between the nitrogen vial headspace and the
atmosphere. At 15,000 feet above sea level in the cargo sec-
tion of a plane, the pressure outside the vial will be reduced,
causing the differential pressure across the package seal to
be about 6 psig. Under these conditions, oxygen will con-
tinue to diffuse into the vial; however, oxygen ow into the
container will be restricted by the net positive pressure inside
the vial, which will convectively force nitrogen and liquid
product out.
Because convective ux is generally much greater than
diffusional ux, leakage theory generally assumes a single-
component system for which only convective ux occurs. In
other words, only a total pressure gradient exists, and there
is no concentration gradient of gas across the seal. One exam-
ple of a parenteral package that may require the consideration
of diffusional leakage is an inert gas ushed vial in which con-
vective ux may approach zero, but enough diffusional leak-
290 Chapter 4
age across the seal could result in the loss of headspace integ-
rity. Another example is a wet-dry syringe, for which a rise in
moisture level of the lyophilized powder could occur due to wa-
ter vapor diffusional leakage moving from the wet chamber
into the dry one. In addition, moisture trapped in the elasto-
mer separating the chamber may permeate out of the elasto-
mer, adding to the cakes total water content.
In conclusion, when considering the microbial barrier
properties of a package, only leakage, not permeation, is a con-
sideration. However, when considering the overall integrity of
a parenteral package system, it may be necessary to consider
the sum of leakage and permeation. For instance, if a product
contains a volatile or solvent component that can sorb into the
container, then permeation plus leakage needs to be consid-
ered. Or, if a parenteral package must maintain an inert gas
or partial pressure headspace, both leakage and permeation
may play a role in satisfactory package performance.
Leakage Units of Measure
Leakage is mathematically dened as the rate at which a unit
of gas mass (or volume) ows into or out of the leak under
specic conditions of temperature and pressure. For example,
a carbonated beverage can may leak 10 cc of carbon dioxide
in 3 months at 60 psig, or a submerged package may leak two
bubbles per second of 1/8 inch diameter when pressurized to
40 psig.
The units of measure commonly used in many literature
references to specify leakage rate are standard cubic centime-
ters per second (std cm
3
/s or std cc/s). According to the interna-
tional metric system of units (SI nomenclature), leakage is
measured in pascal cubic meters per second (Pa m
3
/s). In both
expressions, units of gas mass (std cc and Pa m
3
) indicate the
quantity of gas (air) contained in a unit of volume at sea level
atmospheric pressure (101 kPa). For very precise measure-
ments, standard temperature of 20C (293K) is also specied.
Unless temperature varies widely during an experiment,
Table 4.1 Leakage Units of Measure
Pa m
3
/s
a
Std cm
3
/s
b, c
Std L/day
c
Air at 0C kg/year
1 10 864 400
0.01 0.1 8.6 4
10
4
10
3
86 10
3
4 10
2
10
6
10
5
86 10
5
4 10
4
10
8
10
7
86 10
7
4 10
6
Other units:
1 microliter per second (L/s) 1.33 10
4
Pa m
3
/s
1 microcubic foot per hour (ft
3
/hr) 1.0 10
6
Pa m
3
/s
1 torr liter per second (torr L/s) 0.133 Pa m
3
/s
a
SI nomenclature.
b
More commonly used nomenclature. Also abbreviated as std cc/s.
c
Expressed as quantity of gas (air) in a unit of volume at sea level atmo-
spheric pressure.
however, small changes due to temperature variation are in-
signicant compared to the potentially large differences in
gas pressure. Today, std cc/s is the more common unit of mea-
sure, but because both units of measure are conventionally
used, leakage rates given in this chapter are presented
either in units of std cc/s or Pa m
3
/s, depending on the pref-
erence of the referenced resource. To convert to std cc/s from
Pa m
3
/s, the SI units should be multiplied by a factor of 9.87
or approximately 10. These and other common leakage units
of measure are summarized in Table 4.1 (11).
Leakage Modes and Flow Rates
There are three modes of convective ux leakage that describe
the owpatterns demonstrated by leaking gas. Turbulent ow
is very rapid leakage, followed by slower laminar ow (LF),
and nally the even slower molecular or Knudsen ow. For
capillary pores larger than about 10
4
cm in diameter, gas
leakage is typically turbulent. Turbulent ow rates measure
greater than 10
2
std cc/s (9,12). Laminar ow occurs for capil-
292 Chapter 4
Fig. 4.2 Leak rate convective ux. (Courtesy of the Parenteral
Drug Association, Inc., Bethesda, MD.) (Source: Ref. 9.)
laries about 10
4
cm in diameter, and measured leakage rates
are approximately 10
3
to 10
6
std cc/s (9,10,1214).
Molecular ow is most probable with leakage rates below
10
5
std cc/s (9,12). It is frequently seen in situations of very
low gas pressure as well. Molecular ow leakage is so slow
that it only describes leakage of gases and not liquids. The
mathematical relationships between leakage ux and the dif-
ferential pressure across the seal are described in Eqs. 4
through 6 and are illustrated in Fig. 4.2. More detailed equa-
tions can be found in the references cited above.
Molecular: Q f (P
1
P
2
)
most probably 10
5
std cc/s
(4)
Laminar:
most probably 10
3
to 10
6
std cc/s
(5)
Turbulent:
most probably 10
2
std cc/s
(6)
where
Q convective ux (mass/time)
P
1
upstream pressure
P
2
downstream pressure
Leak Free
When package integrity is critical for acceptable product per-
formance, leakage is generally compared to a leak rate speci-
cation. For example, a pickup truck is leak tight if the truck
bed keeps the smallest nugget of gravel on board. Apacemaker
will leak during its implant life if it exhibits a leakage rate
of greater than 10
9
std cc/s. Vacuum vessels meet a leak rate
specication of no greater than 10
5
std cc/s (15).
Figure 4.3 illustrates approximate gas leakage rates and
Fig. 4.3 Leak rate as a continuum.
294 Chapter 4
their practical signicance. This illustration points out that
leakage is a rate and is therefore a continuum. The practical
signicance of a given leakage rate will depend on the nature
of the substances contained in the package. Very rapid leak-
age is audible (i.e., the noise from a whistling teapot). As leaks
get smaller, they become imperceptible to the human senses.
As discussed in the next section, aqueous liquids have been
shown to migrate through channel leaks of as small as about
10
8
Pa m
3
/s (10
7
std cc/s) measured by helium mass spec-
trometry. Most gases cannot pass through leaks smaller than
about 10
9
Pa m
3
/s (10
8
std cc/s). Because helium gas can
ow through the smallest holes; it is therefore useful as a
tracer gas for detection of leaks as small as 10
12
Pa m
3
/s
(10
11
std cc/s).
Expressions such as leak free, without leaks, or leak
tight are meaningless when used alone. All package seals and
closures leak to some degree. What is meant by those terms is
that any leaks present are so small that they have no practical
signicance. In other words, leakage is occurring below the
established leak rate specication.
ESTABLISHING LEAK RATE SPECIFICATIONS
Liquid-borne Microbial Leakage Specication
The leak rate specication for any given product/package sys-
tem, also called the critical leak rate, is a function of the pack-
ages required protective and containment properties. For par-
enteral products, packages must minimally prevent the loss of
product and prevent the ingress of microorganisms or external
contaminants. Before establishing such a specication, the
leak size that will permit product loss and microbial ingress
must be known.
In the late 1980s, Morton et al. designed a series of tests
to determine the gaseous, liquid, and microbial barrier proper-
ties of vial/closure compression seal systems (16,17). A simu-
lated vial was sealed at various compression forces using elas-
tomeric closures. The gaseous leakage rate was rst measured
Fig. 4.4 Differential pressure laboratory test unit. Differential
pressure decay measured between test vial side of manifold and ref-
erence side over time. (Courtesy of the Parenteral Drug Association,
Inc., Bethesda, MD.) (Source: Ref. 16.)
across the seal using a laboratory-scale differential pressure
testing apparatus (Fig. 4.4). With this device, various elasto-
meric closures could be applied to the test vial within a range
of compression forces and tested for their ability to affect a
seal. Leakage from the test package was measured by rst
opening both valves 1 and 2, ooding the manifold and test
package with nitrogen to a target pressure of 3 psig, then clos-
ing both valves and measuring any detectable differential
pressure change within the manifold over time. The leakage
rates measurable with this system ranged from 10
3
to 10
7
Pa m
3
/s (or 10
2
to 10
6
std cc/s) at 3 psig differential pres-
sure. Typical results obtained are illustrated in Fig. 4.5, in
which leakage rates for one particular type of polymeric-
coated elastomer are plotted against closure compression (17).
296 Chapter 4
Fig. 4.5 Parenteral vial leakage rate versus percentage of com-
pression of coated closures. Modied polypropylene (N-8) lm-coated
closures were used. (Courtesy of the Parenteral Drug Association,
After characterizing the leakage rates of model test pack-
ages, they were then transferred to a separate test manifold
for evaluating the microbial barrier properties of the seal (Fig.
4.6). The test packages were lled with a saline lactose broth
suspension of Pseudomonas aeruginosa at a minimum concen-
tration of 3 10
8
colony-forming units per milliliter. The lled
vials were inverted so that the nish area was immersed in
sterile saline, then the vials were pressurized to 3 psig for 15
minutes to replicate the conditions of the differential pressure
test and a liquid tracer test (see below). The immersion uid
was checked for the presence of the challenge microorganism
by a lter plate count method. Figure 4.7a illustrates the inci-
dence of microbial leakage for vials sealed at various closure
compression levels superimposed on the gaseous leakage
Fig. 4.6 Laboratory test units for bubble, liquid chemical tracer,
and microbial egress leak test comparisons. (Courtesy of the Paren-
teral Drug Association, Inc., Bethesda, MD.) (Source: Ref. 17.)
rates measured by differential pressure decay. As illustrated,
P. aeruginosa was incapable of passing through vial/closure
compression seals that exhibited gas leakage rates of less
than 10
5
Pa m
3
/s (or 10
4
std cc/s). Interestingly, there were
vials that failed to allow microbial ingress even at gaseous
leakage rates signicantly higher than the critical leakage
cutoff rate.
Later research by Kirsch et al. at the University of Iowa
provided more exhaustive microbial challenge test results that
correlated helium leakage rate to microbial ingress (18,19).
While Morton et al. investigated leakage through a vial/clo-
sure compression seal, Kirsch and colleagues studied leakage
through glass micropipettes of various sizes embedded in the
walls of glass vials (Fig. 4.8). A population of vials containing
holes ranging in size from0.1 to 10 min diameter was ooded
with helium and subsequently tested for helium leak rate us-
ing a mass spectrometry leak rate detector (18). The relation-
ship demonstrated between heliumleak rate and nominal leak
diameter is shown in Fig. 4.9. These same vials were then
lled with sterile media and immersion challenged for 24
hours at 35C with a saline lactose suspension of 10
8
to 10
10
298 Chapter 4
Fig. 4.8 Schematic description of the modied pharmaceutical
vials used as test units for the evaluation of mass spectrometry
based helium leak rate measurements. (Courtesy of the Parenteral
colony-forming units of P. diminuta (Brevundimonas dimi-
nuta) and Escherichia coli (19). Prior to immersion challenge,
the test vials were thermally treated to eliminate airlocks
within the micropipet lumen and establish a liquid path be-
tween the microbial challenge media and the test units con-
tents. After immersion challenge, the test vials were incu-
bated at 35C for an additional 13 days.
The results showed that the probability of microbial in-
gress decreased as hole size and helium leakage rate de-
creased (Fig. 4.10). Even under such extreme challenge condi-
tions, only 3 of 66 test vials with log leak rates less than 4.5
std cc/s failed the microbial challenge, which is consistent with
the vial/closure interface leakage results reported by Morton
et al. The probability of microbial ingress dramatically
dropped from over 60% to about 10% within the helium log
Fig. 4.7 Parenteral vial comparative leakage test sensitivity re-
sults. (a) Microbial leakage versus differential pressure gas leakage;
(b) liquid chemical tracer and bubble test leakage versus differential
pressure gas leakage. (Courtesy of the Parenteral Drug Association,
300 Chapter 4
Fig. 4.9 The relationship between nominal leak size and the abso-
lute leak rate. The absolute leak rate was determined using test
units that contained a headspace composed only of helium. (Cour-
tesy of the Parenteral Drug Association, Inc., Bethesda, MD.)
(Source: Ref. 18.)
leak rate range of log 3.8 to 4.5 std cc/s, which roughly
corresponds to a leak diameter range of 0.4 to 1.0 m. Ingress
no longer occurred in this study through holes with helium
leakage rates between 10
5
and 10
5.8
std cc/s. Thus, the work
by Kirsch and colleagues, backed by the results of Morton et
al., is often cited to support a critical leak rate specication of
anywhere from 10
5
to 10
5.8
std cc/s helium leak rate mea-
sured at 1 atmosphere differential pressure and standard tem-
perature conditions for rigid, nonporous parenteral packages.
Liquid Leakage Specication
As stated, parenteral packages also need to prevent the loss
of product or the ingress of external contaminates. Over two
decades ago, a study was performed using metal cans in which
microbial ingress was measured and correlated to water in-
Fig. 4.10 The correlation of microbial failure rate (%) and the
mean logarithm of the absolute leak rate and nominal leak diameter
for modied small volume parenterals. The absolute leak rate (stan-
dard cubic centimeters per second) was determined by mass spec-
trometrybased helium leak rate detection. Microbial failure was
measured by microbial ingress after 24-hour immersion in a bath
(37C) containing 10
8
to 10
10
P. diminuta and E. coli organisms/milli-
liter and a 13-day, 35C incubation. (Courtesy of the Parenteral
gress and helium gas leakage rate (20). Water ingress was de-
termined using a manganese tracer added to the microbial
challenge solution. In this case, water ingress was directly re-
lated to the ingress of microorganisms, which in turn was re-
lated to the log of the helium gas leakage rate measured
through the cans.
This research sparked a similar study by Morton and col-
leagues, who investigated the ability of their model packages
302 Chapter 4
vial/closure seal to prevent liquid leakage. The liquid chal-
lenge consisted of a copper sulfate tracer solution detectable
by atomic absorption spectrometry assay (17). After lling the
simulated vials with tracer solution, the nish areas of the
vials were lowered into 10 ml of water, and a differential pres-
sure of 3 psig was applied to accelerate potential leakage for
a period of 15 minutes (Fig. 4.6). The test was calculated to
be capable of detecting as little as 0.1 l of copper solution in
the water. The ability of this test to detect leakage was com-
pared to that of a bubble test, the previously cited microbial
challenge test, and differential pressure change leakage test.
As seen in 4.7b, copper tracer solution was detected in all con-
tainers known to exhibit gaseous leakage rates greater than
or equal to 10
5
Pa m
3
/s (10
4
std cc/s). This is consistent with
older literature references, which state that aqueous leakage
will not occur when dry air, at the same pressure, leaks at a
rate as great at 10
5
Pa m
3
/s (21). Interestingly, liquid leak-
age detected with copper tracer was found to be a more reliable
predictor of leakage than microbial challenge tests as copper
was found in all vials with leaks above the critical gaseous
rate, but not all vials above this leakage rate exhibited micro-
bial leakage (contrasting graphs a and b of Fig. 4.7).
This research was again collaborated by Kirsch and col-
leagues using test vials with micropipets. As mentioned, prior
to challenging the test vials with microbial suspension, the
vials were thermally treated to ensure a liquid pathway
through the micropipet. Adding magnesium tracer ion to the
challenge solution and checking for its presence inside the test
vials at the end of the incubation period by atomic absorption
spectrometry veried the presence of liquid in the pipet leak
path. This body of data provided an interesting correlation
among helium leak rate, the probability of liquid leakage, and
the chance of microbial ingress, as summarized in Ref. 22. In
the test vial population, all vials demonstrating microbial in-
gress also contained magnesium ion tracer. However, there
were a signicant number of vials that allowed passage of
magnesiumtracer but did not allowmicrobial ingress. Logisti-
cal regression models were used to describe the probability of
microbial or liquid tracer presence as a function of the loga-
rithm of the helium leak rates (Fig. 4.11). The probability of
liquid tracer ingress is greater than microbial ingress at leak-
age rates less than log 2 std cc/s, with liquid tracer detected
for a few leaks as small as about log 7 std cc/s. The probabil-
ity of liquid ingress decreased dramatically for leaks smaller
than about log 5.4 std cc/s. These results support the con-
tention that liquid penetration of a leak precedes and is a re-
quirement for microbial ingress. Again, as with the research
by Morton et al., microbial ingress is not a certainty even when
a rather signicant liquid pathway is present. It can be con-
cluded that sufciently sensitive liquid tracer tests can be a
better predictor of a packages liquid and microbial barrier
properties than microbial immersion challenge tests.
Fig. 4.11 Logistical regression models describing the probability
of microbial or liquid tracer (magnesium ion) as a function of the
logarithm of the helium leak rates. (Courtesy of the Parenteral Drug
304 Chapter 4
Gaseous Leakage Specication Limit
A container that must maintain a specied inert atmosphere
content or vacuum level will most likely require a stricter leak
rate specication. For example, suppose a parenteral vial with
a headspace volume of 5 cm
3
is stoppered under an initial
headspace vacuum level of 5 psia with a maximum allowable
vacuum level of 10 psia at the end of a 2-year expiry.
If vacuum loss were to occur within this allowable range
in an essentially linear fashion, the allowable leak rate could
be simply calculated as
Leak rate Mass/Time or (Pressure Volume)/Time
Leak rate (10 psia 5 psia)(5 cm
3
)/2 years
Converting to SI units,
5 psia 34,470 Pascals
5 cm
3
5 10
6
m
3
2 years 6.3 10
7
s
Therefore, the maximum allowable leak rate is
(3.45 10
4
Pa)(5 10
6
m
3
)/(6.3 10
7
s) or,
2.7 10
9
Pa m
3
/s
It should be noted, however, that if vacuum loss were allowed
to proceed to atmospheric conditions, the leakage rate over
time would be more sigmoidal than linear, with a tapering off
of leakage rate as the vacuum in the vial approaches atmo-
spheric conditions.
In the above example, the maximum acceptable leakage
rate is dened according to SI units of Pa m
3
/s. Alternatively,
leakage can also be dened according to a product-specic end
point. For example, a lyophilized product may contain an ini-
tial moisture level of 0.1% water, with a maximum allowable
moisture level at expiry of 4.0%. If the mass of the lyophilized
cake is 2.0 g, then the maximum allowable leak rate can be
calculated as
Leak rate (Mass of water pickup)/(Expiry dating)
(4.0% 0.1%) (2.0 g)/2 years
39 mg of water per year
Again, this assumes that the maximumallowable limit for wa-
ter content is not near the products saturation point, and that
moisture sorption by the lyophilized solid is occurring in an
essentially linear fashion. If this is not the case, then the
amount allowed over time would need to be calculated based
on the products characteristic moisture sorption isotherms.
Leak Rate Specication Test Conditions
Leakage rate specications should be dened according to
clearly stated test conditions of time, pressure, and tempera-
ture. For pharmaceutical package systems, it is important to
understand that more than one set of standard conditions may
apply to even the same product-package. For instance, a virgin
multidose parenteral vial is required to prevent all liquid-
borne microbial ingress at all conditions of differential pres-
sure anticipated for the products life over extended periods of
time. However, after being punctured at point of use, this
same package cannot be expected to survive the same extreme
challenge conditions. Such a product would be required to pass
different leak rate specications post puncturing. A similar
example is that of a package-device combination system, such
as a vial designed with a spiking device intended to allowprod-
uct access and transfer to an intravenous administration set.
Prior to spiking, the vial/closure seal would be required to sur-
vive liquid microbial challenges without any product loss over
extended times and under all differential pressures represen-
tative of product processes or the distribution cycle. After spik-
ing, the package would need to prevent the ingress of airborne
microorganisms and ensure complete product transfer. Thus,
two very different types of tests and test limit specications
would be needed for the same vial/closure seal. In addition, a
packages seal or closure system may need to be evaluated in
different ways to ensure that processes do not compromise the
306 Chapter 4
package integrity. For example, a vial may need to be evalu-
ated for its integrity during a terminal steam sterilization cy-
cle, as well as during normal shipping and handling. Finally,
different tests and test specications may be needed to dene
the integrity of different seals within the same package sys-
tem. For instance, an induction seal on a liquid-lled bottle
may need different specications and tests than the screw-
thread cap covering the bottle nish.
Leak Rate Specication Terminology and Safety Factors
The terms used to dene the leak rate specication will differ
according to the type of test method used. In some cases, leak-
age is measured qualitatively, and the specication can be de-
ned in absolute terms. For instance, for a package that is
meant to prevent liquid loss and microbial ingress, any detect-
able liquid leakage under the stated conditions of storage and
use would be considered unacceptable. Then, a liquid or micro-
bial ingress leak test specication may be dened as either it
does occur (according to the stated conditions) and the package
is therefore unacceptable, or it does not occur and the package
is considered satisfactory.
On the other hand, when leakage is being measured
quantitatively, leakage specications may be dened in nu-
meric terms. This is the case, for example, when measuring
gaseous leakage by helium mass spectrometry, differential
pressure, or vacuumloss. Once the leakage rate corresponding
to product failure has been dened, a leakage rate limit needs
to be established. Given the fact that gaseous leakage rate is
a logarithmic phenomenon, and many leakage tests are accu-
rate to only within 50% of the true value, it is reasonable to
use a safety factor when selecting a leakage rate specication.
Often, a safety factor of 0.1 is recommended (21). Thus, for a
package that must demonstrate 10
5
Pa m
3
/s as the maxi-
mum allowable leak rate, the leak rate specication would be
established as 10
6
Pa m
3
/s. When referring to gaseous leak-
age, the set of standard conditions most commonly accepted
is that of dry air at 25C for a pressure differential between
1 standard atmosphere and a vacuum (a standard atmosphere
is 101.325 kPa). For practical purposes, the vacuum need be
no better than 1/100 of an atmosphere or 1 kPa. If test condi-
tions are not specied, these standard conditions are generally
assumed (11).
PACKAGE INTEGRITY TEST METHODS
Selection of Equipment
Especially during the development stage of a product-package
system, it is important to demonstrate the ability of the pack-
age to meet the established leak rate specications. Therefore,
a leak test method should be selected that is capable of mea-
suring at or near the specication rate. For quantitative leak-
age tests, the leak test instrument or method should be more
sensitive by at least a factor of 2 than the minimum leakage
to be detected to ensure reliability and reproducibility of mea-
surements. To specify leakage rates lower than necessary to
ensure acceptable package performance, or to select a test
method capable of measuring leakage rates far smaller than
required, is impractical and expensive. The cost of test equip-
ment able to detect a leak of 10
3
std cc/s is negligible com-
pared with that having a sensitivity of 10
12
std cc/s, which
may run 10,000 times higher.
Leak test equipment generally has an optimum window
of performance. For instance, bubble testing by immersion in
water can accurately detect leakage rates of 10
1
to 10
3
std
cc/s. As leaks approach the lower end of this range, a longer
immersion time and perhaps better observation techniques
will be required for detection. Leaks just below 10
4
std cc/s
may be detectable by lengthening the test time, by increasing
the pressure differential, or by adding surfactants to the im-
mersion liquid. This modied method may prove impractical,
however. On the other hand, measuring leaks above 10
1
std
cc/s may require steps to ensure that the package does not
308 Chapter 4
experience too rapid a pressure loss before the package can be
immersed.
There are other considerations in the selection of leak test
equipment, including cost, ease of use, repeatability, testing
speed, and safety. Data integrity, storage, analysis, and dis-
play capabilities are also important criteria. Selection of leak
test methods and equipment should be carefully catered to the
package requirements and the needs of the user.
Test Methods
In this section, various package integrity tests useful for the
evaluation of parenteral package systems are reviewed. These
methods primarily consist of techniques for detecting the pres-
ence of a leak or for measuring the degree of leakage. Methods
that detect the presence of leaks are called qualitative tests
either the package is found to leak or it does not. Such tests
include liquid tracer tests, high-voltage leak detection, micro-
bial challenge tests, or bubble tests. Other tests provide quan-
titative leak results, such as helium mass spectrometry or
some pressure/vacuum decay methods. Then, there are tests
that detect the evidence of leakage, such as near-infrared
(NIR) spectrometry, which measures the presence of excess
moisture in a dry powder product resulting from unacceptable
package integrity. Tests such as acoustic imaging or airborne
ultrasound may help to visualize a possible leak, or they may
serve to better characterize package seals, allowing a more
complete understanding of the mechanisms of package integ-
rity. Other tests measure mechanistic characteristics of a clo-
sure or seal that may relate to package integrity. The residual
seal force test is one example of such an approach.
Each test has advantages to its use and inherent draw-
backs. There is no one perfect test for all situations. In many
cases, more than one test may be needed. Often, more sensi-
tive test methods that characterize several integrity aspects
of a package are required during the earlier development
stages of a product. Later, when the package and critical pro-
duction steps have been dened and controlled, less stringent
test methods may be more appropriate.
The lists of tests discussed and the suppliers mentioned
in this section are not intended to be exhaustive. These are
some of the most common methods, but there are likely nu-
merous others in use. The suppliers are those with whom we
are most familiar. Ultimately, test methods and instrument
supplier selection should be based on a rational understanding
of the package design, the critical nature of any potential
leaks, and the capabilities of the instrument. The nal criteria
for selection are ultimately based on the ability of a test
method and instrument to reproducibly and reliably identify
package integrity failures.
Asummary of package integrity test methods; their levels
of sensitivity; advantages, and disadvantages; reported usage;
and equipment suppliers are presented in Table 4.2. Addi-
tional comments on each of these methods are presented at
the end of this section.
Acoustic Imaging
Acoustic imaging is a technology used to examine hidden or
obscured defects in objects as varied as microchips, construc-
tion joints, ceramics, composites, and laminates. In acoustic
imaging, specially designed piezoelectric transducers convert
electrical impulses into very short ultrasound pulses at fre-
quencies up to several hundred megahertz. An acoustic lens
focuses these pulses to a pinpoint that has the power to dene
tiny features within a sample. After an ultrasound pulse is
sent into the sample, the return echoes are detected by a re-
ceiver. These arrive at different times, depending on the depth
of the feature. A digital waveform on the user interface screen
displays the arriving echoes. This echo-time display can be
programmed to produce planar images or images representing
varying depths within the sample.
This technology has been advertised as a research-and-
development tool for evaluating package seals and closure sys-
tems of pharmaceutical packaging. While it is a very powerful
3
1
0
C
h
a
p
t
e
r
4
Table 4.2 Package Integrity Test Methods
ACOUSTIC IMAGING
Method: Ultrasonic energy is focused into sample sub- Sensitivity: Qualitative imagery
merged in water or other solvent. Echo patterns pro-
duce images of package material interior.
Pros: Cons:
Very sharp image produced Expensive
Structural defects visible, such as channels, delamin- Sample must be immersed
ations Slow
Sophisticated tool for package investigations and develop- Expertise required
ment Not as useful for porous materials
Critical test parameters: Reported usage:
Package structural design Microchip technology, forensic science, con-
Instrument capabilities (transmitters, receivers, imagery struction materials, packages and devices
analysis)
Suppliers:
Sonoscan Inc., www.sonoscan.com
BUBBLE TEST
Method 1: Submerge package in liquid, apply differential Sensitivity: 10
5
std cc/s
pressure, observe for bubbles. Method dependent; qualitative
Method 2: Apply surfactant, draw vacuum, and look for
foaming.
Pros: Cons:
Simple Relatively insensitive
Inexpensive Operator dependent
Location of leaks can be observed Wets package seal; destructive
Good early research or troubleshooting technique Requires gas headspace to be present at leak-
Critical test parameters: age site
Visibility of submerged package (lightning, background,
magnication)
P
a
c
k
a
g
e
I
n
t
e
g
r
i
t
y
T
e
s
t
i
n
g
3
1
1
Time allowed for inspection
Differential pressure across seal
Submersion uid or surfactant surface tension
Operator training and ability
Gas headspace present at leakage site
Package cleanliness
Suppliers: Reported usage:
None Pipes, large equipment, aerosols (warm water
bath test)
GAS TRACER DETECTION
Method: Test tracer gas is used to measure leakage/perme- Sensitivity: Time/instrument dependent
ation across a package seal. Gas is detected either by a Quantitative
coulometric detector (O
2
) or by a photoelectric sensor
(CO
2
or H
2
O). Instruments that invasively test package
headspace for O
2
or CO
2
are another type of gas detec-
tion test method.
Pros: Cons:
Directly correlates to package performance, protection Slow
Potentially highly sensitive Some tests are destructive
Provides total leakage and permeation information
Tracer gas detection technology sensitivity and accuracy Screw-cap bottles, food and beverage con-
Headspace sampling technique (for invasive testing) tainers
Package headspace volume Blister packages, polymer and foil pouches
Concentration of tracer gas inside package
Suppliers:
MOCON/Modern Control, Inc., www.mocon.com
3
1
2
C
h
a
p
t
e
r
4
Table 4.2 Continued
HELIUM MASS
SPECTROMETRY
Method: Helium is used as a tracer gas for detection and Sensitivity: 10
11
std cc/s
measurement of leakage using a mass spectrometer. In- Quantitative
side-out method or sniffer probe methods are two op-
tions when helium is inside the package.
Pros: Cons:
Inert gas tracer May confuse helium diffusion with leakage
Extremely sensitive test Helium must be added
Rapid test time May be destructive
Correlated to microbial and liquid leakage Bombing takes time
Critical test parameters: Expertise required
Helium concentration inside the package May not detect large leaks
Helium sorption onto the package Expensive
Helium background within the test system
Helium background in surrounding test environment
Test time duration
Test chamber size
Test chamber vacuum level
Suppliers: Reported usage:
Alcatel Vacuum Technology, www.alcatelvacuum.com Pharmaceutical packaging, refrigeration
Incon, Inc., www.incon.com units, automotive parts, pacemakers, food
Varian, Inc., www.varianinc.com and beverage containers, drums
P
a
c
k
a
g
e
I
n
t
e
g
r
i
t
y
T
e
s
t
i
n
g
3
1
3
HIGH-VOLTAGE LEAK
DETECTION (HVLD)
Method: High frequency, high voltage is applied to sealed Sensitivity: Shown to detect 5-m holes. Re-
container. Increase in conductivity correlated to pres- ported by suppliers to detect 0.5-m defects
ence of liquid near detectors.
Pros: Cons:
100% Automatic inspection Difcult to validate with standard defects
Clean Liquid-ll product required
Nondestructive
Rapid
Conductivity of package and product Glass and plastic ampules or blow/ll/seal
Product proximity to leak containers; glass vials, syringes
Package cleanliness
Humidity of testing site
Voltage level
Gain setting
Suppliers:
Rommelag, www.rommelag.com
Nikka Densok Limited, www.NikkaDensok.com
LIQUID TRACER TESTS
Method: Package is immersed in solution of a tracer chemi- Sensitivity: 10
5
std cc/s depends on method,
cal or dye. tracer, and detection
Pressure/vacuum or temperature cycling is used to im- Semiquantitative
prove sensitivity.
Leakage is detected visually (dye) or instrumentally (dye
or chemical).
Pros: Cons:
Correlates to liquid leakage and microbial ingress Destructive
Operator independent (instrumental methods) Human variability (visible dye)
Inexpensive Probabilistic so larger sample numbers
Simple to perform needed
Slow
3
1
4
C
h
a
p
t
e
r
4
Table 4.2 Continued
Critical test parameters for both dye and chemical tests: Reported usage:
Immersion exposure time, differential pressure, tempera- All types of packages
ture cycles
Volume of liquid in test package
Tracer solution concentration, viscosity, surface tension
Cleanliness of system and solutions
Critical test parameter for dye only:
Chemical compatibility with product formulation
Human inspection variables (ability, training, environ-
ment)
Suppliers:
Detection instrumentation specic
MICROBIAL CHALLENGE
TESTS
Methods: Containers are media lled, and the seal is chal- Sensitivity: Erratic, method dependent
lenged with microorganisms (in liquid suspension or aero-
sol form). Presence of microbial growth is conrmed visu-
ally or with instrumentation.
Pros: Cons:
May provide direct correlation to microbial integrity Insensitive
No special equipment required Expensive in time, storage, resources
Airborne challenge best approach for tortuous seal tests Slow
Critical test parameters for both immersion and airborne
tests:
Microorganism selected
Concentration of challenge organism
Growth promotion ability of media inside package
Differential pressures, temperatures to maximize ingress
Time of exposure
P
a
c
k
a
g
e
I
n
t
e
g
r
i
t
y
T
e
s
t
i
n
g
3
1
5
Critical test parameters for immersion tests: Reported usage:
Viscosity and surface tension of challenge uid Widely used throughout the pharmaceutical
Presence of liquid path in leak industry
Critical test parameters for airborne tests:
Challenge vessel design and air circulation patterns
Relative humidity and temperature in challenge chamber
Package position and placement in challenge chamber
Suppliers:
None
NONINVASIVE MOISTURE
AND OXYGEN ANALYSIS
Method 1: Moisture. Near infrared spectroscopy. Used to Sensitivity: 0.2% water content (0.5 mg) in
measure powder moisture inside an unopened glass methacholine chloride reported. Calibrated
package. against known standards.
Pros: Cons:
Nondestructive to product or package Unknown
Rapid (less than 1 minute)
Sensitive to trace moisture
Simple to perform
Moisture content range within test method sensitivity Lyophilized and dry powder pharmaceuticals,
Uniformity of test sample moisture validation of terminal sterilization cycle
Suppliers: package integrity
FOSS NIRSystems, www.foss-nirsystems.com
Method 2: Oxygen and moisture. Tunable diode laser spec-
troscopy. Laser light is passed through package head-
space. Frequency of light matched to oxygen or water.
Absorbed light is proportional to headspace contents.
3
1
6
C
h
a
p
t
e
r
4
Table 4.2 Continued
Pros: O
2
Sensitivity:
Nondestructive to product or package Test range of 1100%
Rapid (1 second) Accuracy 0.5% absolute
Cons: Precision 0.1% absolute
Unknown H
2
O sensitivity:
Critical test parameters: Test range of 1100% RH at 25C to 45C
Unknown Calibration data available against known
Suppliers: standards
Lighthouse Instruments, LLC, www.lighthouse Reported usage:
instruments.com Products sensitive to moisture or oxygen head-
space content
RESIDUAL GAS IONIZATION
TEST
Method: High-voltage, high-frequency eld is applied to Sensitivity: not documented.
glass vials or bottles sealed under vacuum. The eld Qualitative
causes residual gas to glow. Vacuum level is veried by
glow or ionization current.
Pros: Cons:
On-line, nondestructive test Inconsistencies in results unless parameters
Rapid carefully controlled
Vial storage time postsealing Lyophilized product
Voltage and frequency
Positioning of voltage eld and conducting plate
Package headspace volume
Package component ionization potential
Gaseous content of residual headspace
Package component geometry
Suppliers:
Electro-Technic Products, Inc., Chicago, IL
P
a
c
k
a
g
e
I
n
t
e
g
r
i
t
y
T
e
s
t
i
n
g
3
1
7
RESIDUAL SEAL FORCE
(RSF)
Method: Vials sealed with elastomeric closures are com- Sensitivity: Dependent on closures, stress-strain
pressed at a constant rate of strain. Stress-strain defor- instrument
mation curves are generated. Second derivative of the
curve RSF value.
Pros: Cons:
Measures closure force post compression Good RSF will vary with elastomer material
Nondestructive (although plastic cap may be removed) and history
No human error
Qualitative measurements
Simple to perform
Elastomeric formulation, size, history (age, sterilization) Parenteral vials (13 to 28 mm)
Design of cap anvil
Rate of strain
Load cell sensitivity
Approach used for detecting RSF second derivative
Suppliers:
Instron, www.dynatup.com
Genesis Machinery Products, Inc., www.genmap.con
ULTRASONIC IMAGING
Method: Ultrasound echoes used to create image of heat Reported usage: Heat seals
seals.
Suppliers:
Packaging Technology & Inspection, LLC, www.ptiu-
sa.com
Because published information on this technology is not
yet available, no further information is offered.
3
1
8
C
h
a
p
t
e
r
4
Table 4.2 Continued
VACUUM/PRESSURE DECAY
Method: Change in pressure or vacuum is measured inside Sensitivity: 10
5
std cc/s
the package (destructive) or outside in a sealed package Time, pressure dependent test.
chamber (nondestructive). Quantitative or qualitative use
Pressure/vacuum change signicantly greater than non-
leaking package is indicative of a reject.
Pros: Cons:
Clean Difcult to detect leaks 5 m
Nondestructive (test chamber method) Some package headspace needed
Relevant to package performance (shipping/distribution)
Relatively inexpensive
Sensitivity good for leaks 5 m
Rapid test (few seconds)
Rigidity and porosity of package Variety of pharmaceutical, food, medical
Headspace content of package device packages
Test chamber geometry, volume
Total volume of internal pressure lines
Initial chamber vacuum/pressure
Time to establish initial chamber conditions
Differential pressure/vacuum decay
Time allowed for pressure/vacuum decay
Suppliers:
Packaging Technologies and Inspection, LLC,
www.packagingtechnologies.com
Wilco AG, www.wilco.com
TM Electronics, Inc., www.tmelectronics.com
P
a
c
k
a
g
e
I
n
t
e
g
r
i
t
y
T
e
s
t
i
n
g
3
1
9
VISUAL INSPECTION
Method: Look for leaks. Sensitivity: 10
1
std cc/s
Pros: Cons:
Simple Insensitive
Inexpensive Operator dependent
Critical test parameters: Qualitative
Quality of inspection area (lighting, magnication, back- Reported usage:
ground) All packages for nal inspection, especially
Time allowed for inspection parenteral packages such as vials, ampules,
Operator training and skill syringes
Product characteristics (color, viscosity, surface tension)
Techniques for leakage acceleration (differential pressure)
Package uniformity and cleanliness
Package seal visibility
Suppliers:
Seidenader Inspection System, www.seidenader.de
WEIGHT CHANGE
Method: Filled, sealed container is stored at various stress Sensitivity: Time dependent, can be excellent
conditions and checked over time for weight loss (liquid
contents) or weight gain (dry contents).
Pros: Cons:
Simple Time consuming test
Directly relates to closure performance Leak location not detected
Quantitative Package material sorption may interfere with
Inexpensive results
Volatility or hygroscopic nature of lled product Vials (including lyophilized products), aero-
Accuracy of scale sols, solid or liquid dosage form packaging
Handling technique
Sorption by package materials themselves
Suppliers:
None
320 Chapter 4
tool that can detect irregularities such as minute defects and
even seal delaminations and weaknesses, the technology re-
quires that the sample be submerged in water or other solvent,
making it a destructive test. It is also relatively slow, with
several seconds to a few minutes required per sample. How-
ever, it does appear to be an excellent tool for investigative
research on package systems when package integrity is in
question.
Bubble Test
Bubble testing is generally performed by submerging the con-
tainer in liquid, applying a differential pressure, and in-
specting for bubbles (Method 1). Alternatively, a surfactant
may be applied to the outer surface of a pressurized package
that can be inspected for foaming (Method 2). Bubble tests de-
termine the presence and location of leaks that allowsufcient
gaseous ow to permit the formation of detectable bubbles in
a submersion uid or in an applied surfactant. The sensitivity
of bubble tests may vary anywhere from 10
1
to 10
5
std cc/s
depending on the length of time given for observation, the
differential pressure applied, lighting, background contrast,
and the surface tension of the immersion liquid. To further
improve method reliability and sensitivity, ultrasound detec-
tion equipment has been advertised as a tool to detect the
foaming noise from subvisible surfactant bubbles. The sensi-
tivity of Method 1 is dened as the number and size of bubbles
seen in the submersion uid in a given period of time for the
given test parameters. For Method 2, the sensitivity can be
dened as the smallest volume of foam visible.
A disadvantage of bubble tests for parenteral products is
the need to wet the package seal, making it impractical for
100%leak detection in the production environment, but useful
as a rapid screening test or troubleshooting tool.
In Ref. 17, bubble testing of stoppered parenteral vials
was compared to liquid leakage of a chemical tracer solution
and to quantitative gaseous leakage rates measured by differ-
ential pressure techniques. The bubble test was optimized by
using surfactant in the immersion uid, a dark inspection
background, high-intensity lighting, a 3 magnifying lens, a
differential positive pressure of 3 psi inside the vial, and a
maximum test time for 15 minutes. Under these special condi-
tions, the bubble test was able to detect leakage rates as low
as about 10
6
Pa m
3
/s (10
5
std cc/s) (Fig. 4.7b).
Gas Tracer Detection
Leak test methods based on detection of gaseous molecules are
generally the most sensitive of all leak test methods. Gases
such as oxygen, carbon dioxide, and water vapor are commonly
used for leak detection of pharmaceutical, food, and cosmetic
packages. The sensitivity of all tracer gas detection methods
generally depends on the test instrument design and capabil-
ity and the allotted test time. The type of gas detection system
selected is generally dependent on which gaseous moiety is
the most critical to package performance. For instance, a car-
bonated beverage container would likely be evaluated using
carbon dioxide test equipment.
Gas tracer methods work by infusing the package with a
tracer gas and placing it in a test chamber through which an
inert carrier gas (e.g., nitrogen) ows across the outside of the
package. Tracer gas migrating out of the package is detected
in the gas carrier gas either by a coulometric detector (O
2
) or
by a photoelectric sensor (CO
2
or H
2
O). Such tests are valuable
in that they quantitatively measure the permeation and leak-
age of a gas that may be critical to the stability of the product.
Thus, products that are sensitive to oxygen, carbon dioxide,
or moisture can be packaged in a system that has demon-
strated its ability to provide adequate protection. Although
the test takes time to allow for gas migration, the results can
be highly sensitive.
Instruments that are designed to pierce a container and
test its headspace for O
2
or CO
2
are another type of gas detec-
tion test method. Although destructive, this approach is useful
for testing oxygen ingress as a function of shelf life for those
pharmaceutical products that require an inert gas headspace.
322 Chapter 4
They may also be used to verify the presence of a critical inert
headspace in packages as part of manufacturing process and
control.
The sensitivity of tracer gas detection devices is based on
the ability of the tester to accurately sample and measure gas
headspace content in the package. Sensitivity is dened in gas
concentration units and can be veried using packages sealed
with reference gas mixtures of known concentration.
Helium Mass Spectrometry
Heliummass spectrometry is a highly sensitive tracer gas leak
detection technique. In fact, helium molecules are so effective
at traversing through pores that heliumdiffusion through ma-
terials may actually be confused with leakage.
When performing helium leak tests, the package is
typically charged with helium, and gas detection occurs
outside the container in a high-vacuum test chamber (inside-
out testing). Sweeping a low-vacuum sniffer probe along
the outside of the package can also detect leaks. Helium may
be added to the package by ooding it prior to sealing, by
injecting helium into the sealed package, or by exposing
the sealed package to an environment of pressurized hel-
ium (bombing). Helium testing is rapid (a few seconds for
inside-out tests) and can accurately detect leaks as small as
10
11
std cc/s.
Kirsch and colleagues, in their investigations of micropi-
pet leaks, selected helium mass spectrometry because of its
sensitivity and accuracy. As discussed, the test vials with mi-
croleaks embedded in the glass walls were ooded with helium
prior to sealing, and heliumleak rates were measured for each
vial (see Establishing Leak Rate Specications). These vials
then served as a model for evaluating the ability of aqueous
liquids and liquid-borne microorganisms to traverse a leak.
Additional studies were also performed in which these helium-
tested vials were used as standards to determine the sensitiv-
ity of two differential pressure testing approaches (see Test
Methods, Vacuum/Pressure Decay).
Sensitivity of a helium leak test is measured in gas ow
rates of standard cubic centimeters per second (std cc/s) or in
helium concentration units of parts per million (ppm). Nation-
ally traceable helium leak standards can be connected to the
tester/package chamber systemto verify its sensitivity and ac-
curacy. The sensitivity of a specic helium test is based on
the ability of the tester to accurately detect helium within the
package system. When testing packages completely ooded
with helium, the mass spectrometer reading is the same as the
package leakage rate. But, if the package contains a diluted
concentration of helium, such as when bombing or injecting
helium into the headspace, the true package leakage rate will
need to be adjusted accordingly.
While helium mass spectrometry is the most sensitive of
all leak test methods, there are drawbacks to its use. Because
the method is so sensitive, any small amounts of helium in
the test system, in the surrounding environment, sorbed onto
the package, or permeating through the package can result
in inaccurate readings. Helium is rapidly lost through larger
leaks, making it difcult at times to detect them. Small pack-
age headspace volumes also make leak detection potentially
more difcult. Also, not all packages can tolerate the level of
vacuum needed for detecting the smallest leaks. Helium must
be introduced into the package, making it a destructive test
for most pharmaceutical products. While the test is very rapid,
bombing processes can take considerable time. And, if the
package headspace is not completely lled with helium, calcu-
lations must be performed to determine the true leakage rate
based on the concentration of helium actually inside the pack-
age. This highly sensitive method is also the most expensive
technique and requires some training to use it appropriately.
However, there is often no substitute for the information this
technology can provide.
High-Voltage Leak Detection
High voltage leak detection (HVLD) is commonly used for con-
tinuous in-line pinhole inspection of liquid-lled packages
324 Chapter 4
formed from a single, electrically nonconductive material,
such as glass ampules or form-ll-seal plastic containers. In
HVLD, a high-frequency voltage is applied to the container.
Any presence of liquid in or near a leak or possibly a thin area
in the wall of the container, will allow increased conductivity
and subsequent rejection of the package (Fig. 4.12). This tech-
nique is very rapid (thousands of units tested per hour), clean,
automated, sensitive (minimumpinhole size claimed is 0.5 m
and is unaffected by package opacity.
One disadvantage of this method is the difculty in quan-
titatively validating the units sensitivity. Often, pinhole de-
tectors are simply adjusted to cull leakers detected by some
other technique, such as the dye immersion test. A creative
way of correlating conductivity to HVLDsensitivity and repro-
ducibility was performed by Sandoz AG using test ampules
Fig. 4.12 Schematic principle of high-voltage leak detection. G
generator, C
1
and C
2
condenser; R
L
resistor. (Courtesy of the
Parenteral Drug Association, Inc., Bethesda, MD.) (Source: Ref. 24.)
made of Teon containing dened electrical resistors (25).
However, this approach does not allow for correlation to hole
size or leakage rate.
More recently, Moll et al. described the validation of an
HVLD method for testing low-density polyethylene blow-ll-
seal containers lled with a hydrogel-based pharmaceutical
product (24). Holes ranging in size from 5 to 30 m were laser
drilled into the walls of the containers. Larger holes of up to
200 m were manually made using a sewing needle. Once the
packages critical testing zones were identied and the appro-
priate positioning of the detectors was determined, test pa-
rameters of voltage and gain were manipulated to establish
the optimum conditions for reject detection with a minimum
of false-positives (Fig. 4.13). The HVLD was able to detect all
pinhole rejects in a population of several hundred test units
over three replicate runs. Difculties reported in this study
include the inability to laser drill holes smaller than 5 m in
the plastic and the complication that exposure to high voltage
changes the size of the holes.
While HVLD is commonly used for 100% inspection of all
glass or plastic containers, this technology is also reportedly
being used to check for defects in stoppered vials and prelled
syringes. No literature references on validation of this leak
detection method for these packages are currently available.
Liquid Tracer Tests
Using a liquid tracer to detect the presence of a leak is a use-
ful, relatively simple, inexpensive, and potentially sensitive
method for detecting defects capable of allowing loss of liquid
product or ingress of liquid-borne microorganisms. But, be-
cause liquid tracer tests are destructive, they are primarily
used for research and development and for production quality
control.
The liquid tracer may be either a dye solution detected
visually or instrumentally with ultraviolet (UV) spectroscopy
or any chemical added to a liquid media detectable by some
appropriately sensitive assay method. When performing a typ-
326 Chapter 4
Fig. 4.13 Blow-ll-seal ampule used in the validation of HVLD.
The shaded areas represent the areas covered by the two detectors
A (at the bottom, or sealing zone) and B (at the top, or head zone).
(Courtesy of the Parenteral Drug Association, Inc., Bethesda, MD.)
(Source: Ref. 24.)
ical liquid tracer test, the liquid productlled container is
immersed in the liquid tracer solution for a selected period of
time. After immersion, the test packages are cleaned and ei-
ther visually inspected for the presence of dye or the contents
are appropriately assayed for tracer chemical. Adding a sur-
factant or a low-viscosity uid to increase tracer solution capil-
lary migration can optimize any liquid tracer test. Liquid in-
gress can also be improved by introducing differential
pressures and/or temperature cycling, including autoclaving,
during tracer solution immersion. Lengthening exposure time
greatly inuences test sensitivity (26). It is also recommended
that the tracer liquid be ltered free of extraneous contami-
nants that might clog any leaks, and that a microbial preser-
vative be added, especially if it is used repeatedly.
Sensitivity of a liquid tracer method is expressed as the
minimumvolume of tracer solution demonstrated to be detect-
able in the packaged product. Because detection is dependent
on the concentration of tracer in the container, sensitivity can
be improved by minimizing the volume of product within the
test container. And, because liquid migration through smaller
leaks is a probabilistic phenomenon, it is important to test a
larger population of test packages when performing liquid
tracer tests to minimize risk of false negative results (see Es-
tablishing Leak Rate Specications, Liquid Leakage Speci-
cation).
When using dyes, it is advisable to select a substance that
is relatively nontoxic, approved for drug use, chemically and
physically color stable in the product to be tested, insensitive
to light and heat, rich in color, nonreactive with the product
or package, soluble in the cleaning solution, and easily dis-
posed of to meet environmental safety requirements. A typical
dye solution formulation may also contain a preservative to
prevent microbial growth. Blue dyes are most commonly used,
although the human eye more easily detects green to yellow-
green. Blue, violet, and red dyes are the colors least easily
seen. Visual detection of dye solutions will give variable re-
sults depending on packaged product visibility, operator abil-
ity, operator training, inspection station design, and inspec-
tion time (12). Therefore, care should be taken before relying
on dye test results generated solely from human inspection.
To eliminate human variability in dye detection, spectrophoto-
metric methods may be used. Jacobus and colleagues reported
a spectrophotometric method for Quinizarin Green dye se-
lected for the leak detection of ampules containing an oil-
based solution for injection (27).
Chemical tracer tests are potentially more sensitive and
semiquantitative than the dye test since detection techniques
such as atomic absorption or high-performance liquid chroma-
328 Chapter 4
tography (HPLC) can be used. As discussed, Morton et al.
tested simulated parenteral vials for liquid leakage using a
copper sulfate tracer solution and an atomic absorption detec-
tion assay (17). Kirsch reported the use of magnesium ion as
a tracer element to verify the presence of liquid pathways in
leaks (22). In both studies, ion tracer test methods were found
to be more sensitive and reliable than microbial immersion
challenge tests (see Establishing Leak Rate Specications,
Liquid Leakage Specication).
Microbial Challenge Tests
Microbial challenge tests are performed by lling containers
with either growth-supporting product or a culture media,
exposing themto microorganisms, incubating the exposed con-
tainers at growth-promoting temperatures, and checking
them for evidence of microbial growth. This approach to val-
idating package integrity is extensively used by the pharma-
ceutical industry in attempts to directly correlate defects to
microbial contamination risk.
Unfortunately, microbial challenge tests may yield er-
ratic results that do not reliably correlate to leak size or pres-
ence. The research to support this was discussed in the sec-
tion, Establishing Leak Rate Specications. Anyone who has
conducted microbial ingress tests knows that they can be te-
dious and difcult to perform. Because it has been shown that
even signicant leak pathways will not always demonstrate
microbial leakage, a large database of samples is needed to
minimize the risk of false-negative results. Whenever possi-
ble, one should consider tests that more accurately predict the
presence of critical liquid leakage, such physical-chemical
package integrity tests or related package assembly and per-
formance assessments. However, there are situations when it
is valuable to determine the ability of a package design or seal
to prevent actual microbial ingress. This is especially true
when attempting to validate the design of a tortuous path bar-
rier intended to prevent the ingress of airborne microorgan-
isms.
There are three basic variations to microbial challenge
tests (9,28). Microbial immersion tests challenge the container
by immersion in an aqueous microbial suspension. Microbial
aerosol tests utilize an aerosolized microbial suspension chal-
lenge. With either the immersion or aerosol tests, a dynamic
stress of vacuum and/or pressure cycling may be used. The
third type of test is called the static ambient test. It involves
the storage of packaged media or product under typical ware-
house conditionsthe stored containers are checked for evi-
dence of microbial ingress over time.
For each of these basic tests, there are many variations
in use, some of which are described more fully below. Test op-
tions should be carefully selected based on the package and
product being tested and on the expected environmental and
processing challenges to be faced by the product/package sys-
tem. For all tests, it is important to include positive control
test packages to demonstrate the ability of the test to detect at
least some of the package leaks of concern. Because microbial
challenge tests are not 100% effective in detecting leaks, the
test package population should be signicantly large to be
sure package integrity problems are identied. Microbial chal-
lenge test method sensitivity is described in terms of the abil-
ity of the test to detect the known positive control samples.
Immersion Tests Very simply, the immersion test in-
volves terminally sterilizing broth- or product-lled packages,
followed by immersion of the entire package or the seal area
in an aqueous suspension of microorganisms for an extended
period of time. Alternatively, packages may be used that have
been aseptically lled with media and checked for growth
prior to microbial challenge (Fig. 4.14) (28). The packages are
then incubated, and nonsterile containers are visually identi-
ed. Often, the microbial identity of any positive samples is
conrmed to screen out any false positives.
In designing an immersion test, it should be remembered
that terminal sterilization may change the nature of some
types of package seals by, for example, relaxation of elasto-
330 Chapter 4
Fig. 4.14 Container/closure system microbiological immersion
test. (Courtesy of Interpharm Press, Buffalo, Grove, IL.) (Source:
Ref. 28.)
mers or distortion of thermoplastic materials. Therefore, if
possible, the sterilization cycle should duplicate the one actu-
ally used for the product. When validating packages intended
for aseptically lled products, it would be best to use asep-
tically lled media test units. Otherwise, it is important to
verify that the seals or closures of the terminally sterilized
test packages have not been impacted by the sterilization
treatment.
An additional test variable to consider is the selection of
the challenge microorganism. Examples of those reportedly
used for such tests include P. aeruginosa (17,28), Pseudomo-
nas diminuta (Brevundimonas diminuta) (19), Enterobacter
coli (19), and Serratia marcescens (29). Size, motility, and via-
bility in the product or culture media are all factors in microor-
ganism selection. To maximize chance of success, the micro-
bial challenge concentration should be at least 10
5
counts per
milliliter at test completion. There are various culture media
that may be used depending on the nature of the challenge
microorganism. Vacuum and/or pressure cycles may be in-
cluded in the challenge test to further stress the package or
to mimic anticipated product processing, distribution, and
storage conditions. Incubation temperatures should be se-
lected that are most appropriate for growth of the challenge
microorganism. In some cases, consideration may be given to
venting the package closure with a sterile needle containing
a vent lter if an obligate aerobic microorganism is used in
the test (28). Microbial growth, as evidenced by cloudiness in
the package, may be detected visually or with instrumenta-
tion. In the case of product-lled packages, verication of non-
sterility may require aseptic ltration and lter plating for
microorganismidentication. Any nonsterile package contam-
inants are generally identied to verify the challenge microor-
ganism as the source of contamination.
As discussed in Establishing Leak Rate Specications,
the sensitivity of microbial immersion challenge tests has
been compared in recent years to that of a bubble test, liquid
tracer tests, helium leakage tests, and pressure decay tests.
These results by researchers working with Kirsch and Morton
all support the following conclusions: (a) liquid-borne micro-
bial ingress is dependent on the presence of a liquid pathway;
(b) ingress is not assured even when liquid pathways are pres-
ent; (c) ingress is a probabilistic phenomenon, with greatest
probability of ingress for larger leaks (greater than about log
5 std cc/s).
Aerosol Tests Packages evaluated by the aerosol test are
placed inside a conned vessel or chamber, in which they are
challenged with a nebulized cloud of microorganisms at con-
centrations of about 10
3
microbes per cubic foot (Fig. 4.15).
Pressure dynamics may be incorporated to simulate antici-
pated product shipping or processing conditions or simply to
optimize microbial ingress. This is a less stringent and less
sensitive test than the immersion test, primarily because
there is no liquid present to provide a means for organism in-
gress, and the concentration of challenge organisms is signi-
cantly less. However, an aerosol challenge may be a more real-
332 Chapter 4
Fig. 4.15 Microbial aerosol test system to evaluate container clo-
sure systems. (Courtesy of Interpharm Press, Buffalo Grove, IL.)
(Source: Ref. 28.)
istic test for those packages that contains no uid and rarely,
if ever, are wetted during manufacture or product life.
Considerations is designing an appropriate aerosol chal-
lenge test include vessel design and package position and
placement in the chamber, all of which can affect the unifor-
mity of the microbial suspension. The environmental condi-
tions of humidity and temperature should be selected to en-
sure the viability of the microorganisms. In some cases, the
surfaces being tested may be coated with agar, broth, or a dilu-
ent such as glycerol to prevent impacted microbes frombecom-
ing desiccated. An excellent review of aerosol challenge tests
along with cited references of original research can be found
in Ref. 28.
Static Ambient Tests Static ambient tests involve placing
the product of media-lled packages in storage and evaluating
them over time for sterility. This type of test may utilize pack-
aged culture media taken from the lling validation run for
an aseptically lled product. Product stored and tested at ex-
piry for sterility is also a type of static ambient test. Actual
or simulated shipping tests of the lled package may be in-
cluded as part of the test. Because there is no concentrated
microbial challenge to the product, this type of test is the least
effective at detecting leakage. Therefore, static ambient tests
are not recommended unless there is a specic justication to
support their use.
Noninvasive Moisture and Oxygen Analysis
Interest has been building in recent years in two relatively
new methods of noninvasive analysis of packaged product
moisture and oxygen content: Near Infra-Red spectroscopy
(NIR) and tunable diode laser spectroscopy (TDLS).
In NIR spectrometry, the NIR spectrometer is operated
in the reectance mode to test the contents of a closed glass
container. Because the water OEH band shows a strong sig-
nal in the NIR region of the electromagnetic spectrum, low
levels of moisture in the package contents can be measured.
The test may be used qualitatively, or the measurements can
be calibrated and used quantitatively. Reported to be rapid
(less than 1 minute per test) and simple to use, this method
shows promise as a tool for evaluating the package integrity
of dry powder and lyophilized products.
In 1993, Last and Prebble reported the use of NIR for
lyophilized product moisture evaluation (30). More recently,
Birrer et al. used NIR as a way to verify the package integrity
of parenteral vials during a terminal steam sterilization cycle.
The test vials were each lled with 250 mg of methacholine
chloride powder prior to steam exposure. At the end of the cy-
cle, this highly hygroscopic powder was visually examined for
evidence of liquefaction and was tested by NIR. NIRwas found
to be more sensitive than visual inspection and was able to
detect as little as 0.5 mg of water, or 0.2% of the total powder
content (31).
TDLS is another, even newer, method for nondestruc-
tively testing packages for relative humidity, as well as oxygen
content in the package headspace. This technology, marketed
334 Chapter 4
by Lighthouse Instruments, LLC, works by shining a laser
light through the container headspace and tuning the laser
frequency to match the internal absorption frequency of the
molecule being analyzed. Readings are calibrated against
known standards for quantitative reliability. It is reported
that either relative humidity or oxygen content in the range
of 1% to 100% can be measured in a single second (32).
Residual Gas Ionization Test
The residual gas ionization test (also called the vacuum ion-
ization test or the spark-coil test) by Electro-Technics Prod-
ucts, Incorporated, has been cited in the literature as a tool
for verifying the presence of vacuum in vials or bottles (33).
A high-voltage, high-frequency eld is applied to the bottles
or vials, which causes ionization of the residual headspace gas.
As ions and electrons recombine, visible light is observed. Ob-
servation can be made manually or with the aid of a conduct-
ing plate located near the bottle wall that measures the level
of ionization current. Measurable current or visible light is in-
dicative of the loss of vacuum within the bottles. The sensitiv-
ity of this method is based on its ability to accurately and reli-
ably detect pressure exceeding a specied acceptance limit in
evacuated packages. Control packages sealed under known
levels of partial pressure should be used for sensitivity and
validation studies. Sensitivity is dened as the range of partial
pressure readings that can be detected as a function of voltage,
frequency, the package, and the test package system congu-
ration. While this is a nondestructive, rapid test that poten-
tially permits 100% batch testing, residual gas ionization does
have its drawbacks. Namely, it is essentially a qualitative test
and care must be taken to control all test parameters to pre-
vent inconsistencies in test results (34,35).
Residual Seal Force
The parenteral vial serves as a useful illustration of howpack-
age assembly and processing variables correlate to package
integrity. Effective sealing of vials is a function of the dimen-
sions of the package components and the capping operation.
Controlling package component dimensions will ensure con-
sistent stack height of the vial plus the closure relative to
the aluminum seal skirt length. A longer aluminum seal than
necessary, or a shorter stacking height, will result in a loosely
crimped vial. Alternatively, a seal skirt too short or a taller
stacking height may result in a poor seal due to closure dim-
pling.
Elastomeric closures crimped to the nish of parenteral
vials are subject to capping machine head pressure and crimp-
ing force, which act to compress and hold the stopper onto the
top of the vial. The compressive force the closure continues to
exert onto the vial nish after sealing ensures package integ-
rity. This force is termed the residual seal force (RSF). Because
closures are viscoelastic in nature, the RSF will decrease as
a function of time, processing procedures, and elastomer com-
position.
RSF values can be determined using a constant rate of
strain stress tester, also called a Universal Tester, such as
manufactured by Instron. Morton and Lordi rst reported the
use of a Universal Tester to measure RSF values in the late
1980s (36). Alternatively, Genesis Machinery Products, Incor-
porated, has recently introduced an Automated Residual Seal
Force Tester (ARSFT) that works according to the same prin-
ciple. To perform an RSF test, a specially designed aluminum
cap is placed on top of the sealed vial and placed on a compres-
sion load cell of the Universal Tester or ARSFT. The vial is
then slowly compressed at a consant rate of strain, and a
stress-deformation response curve is generated. The RSF is
the force at which the slope of the curve demonstrates a notice-
able decrease.
Ludwig et al. worked to optimize the RSF method by mod-
ifying the metal cap anvil that is placed on top of the vial
(37,38). Rounding the top of the metal cap anvil helped to
make a more uniform compression of slightly imperfect vials.
Making the cap t more tightly helped improve centering of
336 Chapter 4
the cap anvil onto the vial (Fig. 4.16). In addition, Ludwig and
coworkers developed a program that automatically collected
the RSF value from the second derivative of the strain-strain
deformation curve. Both 13-mm and 20-mm nish vials were
tested by this technique. Figure 4.17 illustrates an ideal RSF
curve. Stress deformation data produced from ve replicate
readings of a single vial using an optimized cap anvil are
shown in Fig. 4.18.
Morton and Lordi published data to demonstrate the use-
fulness of the RSF method (39). Figure 4.19a shows the range
of RSF values seen among three lots of the same formulation.
Figures 4.19b and 4.19c demonstrate how factors such as time
and terminal steam sterilization can change RSF values. Be-
cause residual seal force is a direct indication of how tightly
a parenteral vial is sealed, it was shown to be possible to corre-
late RSF to leakage rate. First, closures of various elastomeric
materials were applied to test vials at a range of seal forces.
Some of the vials had been etched with channel defects across
the nish. For each closure/compression seal combination, the
leakage rate was measured using a differential pressure tes-
ter, and the residual seal force was measured. A vial was con-
sidered sealed once the vials leakage rate dropped below the
leak detectors lower limit of detection. As illustrated in Fig.
4.20, leakage was correlated to greater defect depth on the vial
nish area, lower closure compression force, and lower RSF.
Different amounts of closure compression was required to seal
a given defect, depending on the viscoelastic nature of the clo-
sure material. (For additional information on this differential
pressure test, see Test Methods, Vacuum/Pressure Decay.)
Finally, a few comments are offered about RSF tech-
niques. It is our experience that RSF tests are best performed
with only the aluminum seal portion of the cap, without any
added plastic top. As demonstrated by Ludwig et al., the strain
rate must be optimized according to the size of the closure,
with 13-mm closures requiring a lower rate of strain than 20-
mm closures. Sensitivity is measured in units of force; the de-
gree of sensitivity will be a function of the accuracy of the in-
Fig. 4.16 Residual seal force technique. Ludwig-modied cap anvil
positioned on parenteral vial. (Courtesy of the Parenteral Drug Asso-
ciation, Inc., Bethesda, MD.) (Source: Ref. 37.)
338 Chapter 4
Fig. 4.17 Stress-deformation plot of smooth load and second deriv-
ative versus time showing the nal output from the RSF determina-
tion macro. The upper and lower tangent lines are drawn, and the
Instron residual seal force (IRSF) and regression correlation coef-
cients are located in the data box. (Courtesy of the Parenteral Drug
strument and the error inherent in stress-strain readings. The
deviation in RSF values can be determined by compiling the
error seen among a population of vials sealed with a given
elastomeric closure formulation, over a range of sealing forces,
among operators, and between instruments. It has been our
experience, supported by research by Ludwig et al., that opti-
mized residual seal force tests can provide very reliable infor-
mation on the quality of parenteral vial compression seals.
Ultrasonic Imaging
Ultrasonic imaging is a new technology currently available
through Packaging Technologies and Inspection, LLC. (PTI).
Fig. 4.18 Stress deformation data produced from ve compres-
sions of a single vial using Ludwig-modied cap anvil C. The test
vial was sealed with a 20-mm diameter Helvoet 6207 lyophilization
closure. Plots were separated on the x-axis for clarity of presenta-
tion. (Courtesy of the Parenteral Drug Association, Inc., Bethesda,
MD.) (Source: Ref. 37.)
This technique is unusual in that ultrasonic imagery of a pack-
age or its materials can be performed without immersing the
package in a conducting medium such as water. Novel trans-
ducers are used to allow extremely high transmission of ultra-
sonic energy through air and other gaseous media. Reected
and transmitted ultrasonic signals are used to generate sur-
face and internal images of a package that allow the detection
of internal discontinuities, the examination of microstruc-
tures, and a comparison of material characteristics. To date,
airborne ultrasound has been found to be useful for testing
heat seals for the presence of contaminants, leaks, and seal
340 Chapter 4
Fig. 4.19 Residual seal force (RSF) test results. (a) RSF closure
lot-to-lot differences; (b) time-after-capping effects on RSF; (c) termi-
nal steam sterilization effects on RSF. (Courtesy of the Parenteral
weaknesses, and for structural density. Once this technology
becomes fully developed, it should be useful as a nondestruc-
tive tool for research, investigations, and production quality
control of a variety of package seal/closure surfaces.
Vacuum/Pressure Decay
Vacuum or pressure decay leak tests are often used for many
package applications, including food, medical device, and
pharmaceutical packaging. One major advantage such tests
offer is they can usually be performed on the packaged product
as is; the package is not contaminated by the addition of dyes,
chemicals, or gases, and it usually is not physically damaged.
The principal behind this technique is to place the package in
a tightly closed test chamber, exert a pressure or vacuum in-
side the chamber, and then measure the rate of pressure/vac-
uum change in the chamber over time. The rate or extent of
change is compared to that previously exhibited by a control,
342 Chapter 4
Fig. 4.20 Leakage versus percentage closure compression on sim-
ulated parenteral vials. Sealing ability of closures compared for vials
without defect versus those with channel defects etched into the vial
nish. (Courtesy of the Parenteral Drug Association, Inc., Bethesda,
MD.) (Source: Ref. 39.)
nonleaking package. Signicantly greater change for a test
package is indicative of a leak.
As described in the section, Establishing Leak Rate
Specications, Liquid-borne Microbial Leakage Specication,
Morton et al. used the concept of pressure change to evaluate
the sealing ability of elastomeric vial closures. A differential
pressure transducer attached to a test manifold was used to
measure the leakage from simulated test vials stoppered with
various elastomeric closures (Fig. 4.4) (16). With this device,
stoppered vial leakage rates between 10
3
and 10
7
Pa m
3
/s
(10
2
and 10
6
std cc/s) were quantitatively measured. The lin-
ear regressions in Fig. 4.20 represent the percentage the clo-
sure ange was compressed as a result of capping the (x-axis),
versus the residual seal force value ( y-axis) for vials stoppered
with closures of differing viscoelastic properties. The win-
dows in this gure indicate the minimum amount of closure
compression required to seal defective test vials. The defects
consisted of channels of differing depths etched across the
sealing surface. A vial was considered sealed once the vials
leakage rate dropped below the leak detectors lower limit of
detection. As illustrated in the gure, leakage was correlated
to greater defect depth on the vial nish area and lower com-
pression force of the closure onto the vial. Different amounts
of closure compression were required to seal a given defect de-
pending on the viscoelastic nature of the closure material. (For
an explanation of residual seal force, see Test Methods, Re-
sidual Seal Force.
One type of vacuum/pressure test is the vacuum reten-
tion test. To perform this test, samples from a population of
packages sealed under vacuum are checked for the presence
of vacuum over time. Vacuum measurement is made using a
vacuum gauge connected to a needle for puncturing the
package.
Today, pressure/vacuum decay instruments are commer-
cially available for nondestructively testing all types of phar-
maceutical packages. Packaging Technologies and Inspection,
LLC, TM Electronics, Incorporated; and Wilco AG all produce
instruments geared toward parenteral package evaluation for
laboratory and production use. In each case, the package is
placed in a test chamber, and a vacuum or positive pressure
is established within the chamber. The test chamber pressure
is monitored for any change resulting from package leakage
within a programmed test time. If the pressure change ex-
ceeds the previously established baseline for nonleaking pack-
ages, the container is rejected. When selecting a test system,
344 Chapter 4
there are several factors that should be considered. The sensi-
tivity and reliability of the instruments differential or abso-
lute pressure transducer and the capabilities of the internal
pressure/vacuum generator are two critical considerations.
The test chamber and all internal pressure lines should be of
minimal volume for maximum pressure change sensitivity,
and they should be well constructed to reduce risk of back-
ground leaks. Chamber designs are available from suppliers
to restrain nonrigid packages, allowing testing of exible
pouches, bags, and lidded cups. Instruments are available that
have a dual-test method option that permits rejection of gross
leaks prior to testing for smaller leaks. Instruments should
also be compared for their ability to offer any diagnostic test
information and data analysis features. The last consideration
is, of course, the ability of the instrument to reproducibly and
reliably detect defects for a given package type.
When validating an instrument for test method sensitiv-
ity, it is important to develop and optimize an appropriate test
method. Several critical parameters can be manipulated in
test method development. The rst is initial chamber target
vacuum or pressure. More rigid, nonporous packages can tol-
erate a greater differential pressure, allowing for a more sensi-
tive method. Other parameters include time to achieve cham-
ber vacuum/pressure, extent of differential pressure/vacuum
change, and time allowed for pressure/vacuum change. Ma-
nipulation of these variables will enable the establishment of
the most optimum method for a given package. Testing known
good packages and comparing the results to a population of
leakers then completes instrument sensitivity validation.
Leakage can be accomplished by using actual packages with
typical known defects, by using packages that have been modi-
ed with controlled defects (such as the Wilco research re-
ported below), or by introducing a calibrated standard peak
rate into the instruments test chamber. Sensitivity is dened
as the largest hole, defect, or ow of gas detectable using the
dened test method.
To determine the sensitivity of commercially available
vacuum/pressure decay instruments, two published research
studies were performed using instruments manufactured by
Wilco AG. The rst study tested for leakage from empty vials
(40), and the second used solvent-lled vials (41), so that pres-
sure changes resulting from either leakage of package head-
space gases or from vaporization of package liquid contents
could be measured. In both studies, test vials with micropipet
leaks in the glass walls were used, ranging from 0.1 to 10.0
m in diameter. These test vials were identical to those em-
ployed by Kirsch and colleagues from helium, liquid ingress,
and microbial challenge leak tests (see Fig. 4.8 and Establish-
ing Leak Rate Specications). Each test vial was measured
for quantitative leakage by helium mass spectrometry both
before and after differential pressure testing to verify the size
of the leak present. For the Wilco Liquid Filled Container
(LFC) test method, pharmaceutically relevant aqueous sol-
vents were used to determine the impact of solution vapor
pressure on leak detection.
Results showed that the standard vacuum decay test var-
ied in its ability to detect leaks depending on the length of
time allowed for vacuum decay. The data in Table 4.3 indicate
that the practical limit of sensitivity for this technique and
Table 4.3 Estimated Limit of Detection for the Vacuum Decay
Test in Terms of Helium Leak Rate and Nominal Pore Diameter
Vacuum
test time Detection limit Corresponding nominal pore
(s) (std cc/s) diameter (m)
5 1.1 10
3
2
10 8.8 10
4
2
60 4.2 10
4
1
900 6.6 10
5
0.6
Source: Ref. 40.
346 Chapter 4
instrument using empty, helium-lled vials is between log 3
and 4 std cc/s (40).
The LFC results showed about a 14-fold improvement in
leak detection sensitivity. For the LFC method, the practical
minimum detectable leak was associated with a leak rate of
5 10
4
to 6 10
4
std cc/s, which corresponds to a leak size
of between 1 and 2 m. Figure 4.21 shows a direct correlation
among LFC, standard vacuum decay leak test methods, and
helium leak rates. Interestingly, the relationship between
LFC response and leak rate did not vary greatly with solvent.
Because both studies used vials that had also been tested
for helium leak rate, and because helium leak rate for these
packages had been previously correlated to the probability of
liquid microbial ingress (see Ref. 19 and Establishing Leak
Rate Specications), an indirect correlation could therefore
be made between these vacuum decay methods and microbial
ingress. Figure 4.22 illustrates the relationship between the
probability of microbial ingress and LFC measurements. The
Fig. 4.21 Direct correlation between Wilco LFC, standard vacuum
decay leak test methods, and helium leak rates. (Courtesy of the
Parenteral Drug Association, Inc., Bethesda, MD.) (Source: Ref. 41.)
Fig. 4.22 Derived relationship between the probability of micro-
bial ingress and Wilco LFC vacuum decay leak test method. (Cour-
tesy of the Parenteral Drug Association, Inc., Bethesda, MD.)
(Source: Ref. 41.)
effective detection range for the LFC method (i.e., when the
ratio of test unit DP to intact unit DP is signicantly greater
than 1) corresponds to a leak size region associated with 40%
to 100% chance of microbial ingress.
In conclusion, because of the time factor required for
package testing, todays vacuum/pressure decay instruments
can serve as a reliable check for larger leaks (greater than
about 10 m) when used in support of production line opera-
tions. However, when test time can be extended, such as in
research and development or during quality control testing,
detection of smaller leaks (less than 2 m) is certainly pos-
sible.
Visual Inspection
Visual inspection is certainly the easiest leak test method to
perform, but it is also the least sensitive. Visual inspection is
essentially a qualitative method useful for detecting only
348 Chapter 4
larger leaks. Visual inspection can be improved with the use
of magnication and special lighting and background.
Exposing the package to differential pressure is another way
to accelerate potential leakage. When attempting to validate
such a method, sensitivity is dened as the incidence of fail-
ures detected in a mixed population of positive control leakers
as well as good packages. Positive controls should include
leaks with sizes as close as possible to the smallest leaks of
concern thought to be detectable by this approach. Because
visible inspection techniques are considered relatively unreli-
able and insensitive, they are not recommended as the sole
leak detection method for any package application. However,
despite all precautions, package failures do occur that are visi-
bly evident, making visual inspection of products and pack-
ages a useful part of the inspection and testing of any product-
package system.
Weight Change
To perform weight change tests, a population of lled and
sealed containers is monitored for weight loss (for liquid-lled
packages) or weight gain (for dry-lled packages) over time.
Test conditions may include container storage at extremes of
relative humidity and/or temperature. Weight change tests
are required package integrity test methods in the U.S. Phar-
macopeia (USP) for unit-dose and multiple-dose packages for
solid oral dosage forms, as well as for bottles intended for liq-
uids (42,43). Weight gain/loss tests are quite accurate and
simple to perform and provide useful information that directly
correlates to packaged product performance.
Test method sensitivity is dened as the smallest weight
change that can be reliably detected given the sensitivity of
the scale, the weight and the moisture sorption tendencies of
the package, and the variability of the testing and handling
techniques used. In some situations, control packages con-
taining inert materials are useful comparisons for identifying
signicant weight changes in test packages. It is for this rea-
son that the USP method for testing plastic bottles for weight
gain includes a comparison of desiccant-lled bottles to glass
beadlled bottles (42).
Daukas and Trappler reported test results in 1999 that
showed how vacuum loss in lyophilized vials can be quantita-
tively and nondestructively measured quite accurately by
monitoring the weight gain of vials after sealing (35). Slow
leakage and loss of vacuum were evidenced by a signicantly
greater weight gain for compromised vials compared to control
vials. A correlation in results was demonstrated for vials
tested by both weight gain and by a residual gas ionization
technique (by Electro-Technic Products, Inc.).
One drawback of weight change tests is the extended time
required to allow for the change to occur. Weight loss studies
can be expedited by storing the packages at elevated tempera-
tures or low humidity or by lling the containers with materi-
als that more rapidly lose weight. For instance, dry ice has
been used in weight loss tests of intravenous screw-cap bottles
(44), and isopropanol has been lled into parenteral vials stop-
pered with closures at various compression levels (9). Con-
versely, weight gain of dry solids or powders, including ly-
ophilized products, can be accelerated by exposure to
high-humidity conditions. In such cases, the package being
tested should be allowed to equilibrate to ambient tempera-
ture in a dry atmosphere prior to reweighing to prevent error
in weight determinations from moisture sorption to the pack-
aging materials themselves.
THE CHANGING PHARMACEUTICAL INDUSTRY
The pharmaceutical industry has grown familiar with tradi-
tional parenteral dosage from package systems. Glass am-
pules, stoppered glass vials or bottles, plastic ophthalmic solu-
tion dropper-tip bottles, and prelled syringes have been
successfully manufactured for decades. Such simple package
systems are typically thought to have well-understood sealing
mechanisms, so some have felt that the integrity of these sys-
tems has been validated through years of successful manufac-
350 Chapter 4
turing experience. However, every year package recalls con-
tinue to be reported due to package integrity failure and
leakage from even conventional packages.
The continual challenge to provide integral packaging is
now complicated by the growing demand for novel, more com-
plex package systems. Parenteral packages are being designed
for easier preparation and safer administration. The ADD-
Vantage
vial/exible container admixture system by Abbott

Laboratories (Fig. 4.23), the Heparin Dextrose double-bag
product by Baxter Healthcare (Fig. 4.24), and the Vetter Lyo-
Ject
dual chamber syringe designed to contain both the in

situ lyophilized product and diluent (Fig. 4.25) are all early
examples of this approach to packaging. Newer packages now
on the market include the Inter-Vial
and Vari-Vial
systems
by Duoject Medical Systems, Incorporated (Fig. 4.26), the
ClipnJect reconstitution system for use with lyophilized
pharmaceuticals by West Pharmaceutical Services (Fig. 4.27),
the Bio-Set transfer devices by Baxter Healthcare Corpora-
tion (Fig. 4.28), and even form-ll-seal packages that can be
used for unit-dose intravenous line ushes, like the Viringe,
(formerly known as the Vasceze) by Avitro, Inc. (Fig. 4.29).
Not only are newer package designs available, but also pack-
aging materials themselves have continued to evolve. For ex-
ample, elastometric closures can be coated with more inert,
lower particulate shedding polymeric materials, and less reac-
tive, sterilization-compatible plastics have been commercial-
ized for vials, bottles, and exible containers.
Sterilization method changes have also taken place over
the last 10 to 15 years that inuence package selection and
design. The typical techniques for sterilizing either package
components or packaged product using dry heat, steam, ethyl-
ene oxide or gamma irradiation are still widely used. But, ad-
vances in steam sterilization technology allow for pressurized
cycles that can accommodate packages prior to blow-out, as
well as higher temperature/more rapid cycles that permit
sterilization of marginally thermal-stable chemical com-
Fig. 4.23 Abbott ADD-Vantage package system. (Courtesy of Ab-
bott Labs, North Chicago, IL.)
352 Chapter 4
Fig. 4.24 Heparin dextrose double bag. (Courtesy of Baxter
Healthcare Corp., Round Lake, IL.)
Fig. 4.25 Vetter Lyo-ject
by Vetter GmbH. (Courtesy of Vetter

GmbH, Yardley, PA.)
pounds. In addition, E-beam sterilization, hydrogen peroxide
gas, and pulsed light are viable sterilization options.
The ability to accurately test packages for integrity has
improved with the variety of package integrity test methods
available. There is a growing body of literature on parenteral
package integrity that explains the concept of critical leak-
age and that provides information on the proper use of leak
test methods.
Finally, todays regulatory climate requires that, for even
simple package systems, the integrity of the package must be
demonstrated both during development and throughout the
marketed products shelf life. This requires that package in-
tegrity be incorporated in the product development program
354 Chapter 4
Fig. 4.26 Inter-VialPLUS System by Duoject Medical Systems,
Inc. (Courtesy of Duoject Medical Systems, Inc., Bromont, Canada.)
and sustained as a component of marketed product stability
studies.
PACKAGE INTEGRITY AND THE PRODUCT LIFE CYCLE
Now that the concepts of leakage have been explained, some
of the package integrity test methods have been described, and
the challenges of newer package designs have been touched
upon, it is time to explore how package integrity ts into the
products life cycle, from early development to marketed prod-
uct shelf life expiry.
Package Selection
To ensure successful product development, it is necessary to
follow a carefully outlined approach to package development.
Package components should be rationally selected based on
their physicochemical compatibility with the product, as well
Fig. 4.27 ClipnJect
by West Pharmaceutical Services, Inc.

(Courtesy of West Pharmaceutical Services, Inc., Lionville, PA.)
as for their proper t and nal package functionality, and for
their ability to withstand processes such as sterilization and
the anticipated product distribution cycle. Critical dimensions
need to be dened and the specications set to ensure satisfac-
tory package performance under worst-case conditions. The
dimensions of the components should be specied according
356 Chapter 4
Fig. 4.28 Bio-Set
Luer Injection System by Baxter Healthcare

Corporation. (Courtesy of Baxter Healthcare Corp., Round Lake, IL.)
Fig. 4.29 Viringe Vascular Access Flush Device, Avitro, Inc.
(Courtesy of Avitro, Inc., Zephyr Cove, Nevada.)
358 Chapter 4
to appropriate t, clearance, and interference. Component
samples representing multiple component lots should be eval-
uated at the dimensional specication limits for their impact
on package functionality and package integrity.
Package Function and Leakage Specications
When designing a parenteral product package, it is important
to include a denition of the packages intended functions. All
parenteral packaging must maintain the sterility of the con-
tents. In addition, the package must contain the liquid or dry
powder product. This may include the maintenance of low lev-
els of moisture or oxygen or low atmospheric pressure in the
package headspace. For multicompartment packaging (for ex-
ample, wet/dry or wet/wet packaging systems), it is important
that the contents of each chamber be kept isolated until the
time of mixing. The package should also function appropri-
ately to deliver product to the patient. For example, a prelled
syringe must break away and extrude with reasonable force
and without leakage of the contents, package/delivery system
combinations must mate without contamination or loss of
product, and a stoppered multidose vial must reseal after
puncturing and not core.
Once these functions have been identied, leak rate spec-
ications can be established. For example, a container in-
tended to simply maintain sterility and prevent the loss of the
liquid contents needs to meet appropriate leak rate specica-
tions that reect liquid leak limits. Containers required to pre-
serve a dened headspace will need to be assigned a tighter
specication based on the allowable headspace uctuations.
More than one specication may be needed to differentiate
between long-term storage performance and point of use func-
tionality. For example, a multidose lyophilized or powder-
lled vial may need to meet rather strict gaseous leakage spec-
ications during its shelf life. Once reconstituted, it is only
important that no liquid leakage occur, and that any microor-
ganisms present in the atmosphere or on the outer package
surface be prevented from entering the system. After leakage
specications have been agreed upon, nished product leak
test methods can be selected to reect these specications.
Leak test methods should optimally include physical tests de-
signed to provide rapid, clear, and reliable results for screen-
ing packages throughout the development process.
Package Filling and Assembly
Processing and assembly variables should be understood for
their potential impact on package integrity. During produc-
tion, the product is lled into the package and the package is
closed. This assembly process may include, for example, torqu-
ing a cap onto an ophthalmic bottle, crimping an aluminum
seal onto a stoppered vial, or ame-sealing a glass ampule.
Each process entails several variables that ideally should be
evaluated at the proposed operational limits for their impact
on package integrity. Each of these variables should be moni-
tored by some physical or mechanical means to ensure that
the assembly process is kept under control. For example,
ame-sealed glass ampules need to be tested for integrity
when sealed at the limits of gas and oxygen ow rate, lling
speed, ampule seal neck target location, and product ll vol-
ume. Screw-cap containers need to be tested for cap removal
torque and leakage over time as a function of initial capping
torque force. To assure adequate package integrity, it is useful
to assess components made at the dimensional specication
limits and to compare multiple component lots for their di-
mensional and functional consistency.
Components tested should include some that have been
processed and sterilized according to worst-case conditions
known to cause component physical deformation or degrada-
tion. The impact of postassembly processes, such as terminal
sterilization, on package integrity certainly must be evalu-
ated. Sterilization cycle validation should include package in-
tegrity checks under the worst-case conditions of package de-
sign, assembly, and sterilization.
360 Chapter 4
Finally, careless or improper handling of components and
packages have resulted in package integrity failures and
should be prevented through proper procedures and training.
Packaged Product Distribution
The distribution environment challenges of temperature,
pressure, humidity, shock, and vibration should be considered
in evaluating containers for integrity. In todays world of
heightened security, the impact of X-ray exposure should also
be considered. To test for the effects of the distribution envi-
ronment on package integrity, packages may simply be
shipped by appropriate means of transportation. Sophisti-
cated monitoring systems are available that can record the
temperature, humidity, and/or the shock/vibration environ-
ment actually seen by the product during shipment. Such tests
give valuable real-life data; however, it is difcult or impos-
sible to isolate the cause of a reported package integrity fail-
ure. To gain insight into specic integrity test failures, simu-
lated shipping studies can be performed in the laboratory.
Packaged Product Storage
It cannot be assumed that package integrity of rst-manufac-
tured product remains unchanged over its shelf life. Aged
product will often exhibit different package integrity over
time. Frequently, this is due to the changes in inherent visco-
elasticity of polymeric materials used in packages. This is ex-
emplied by the decay in residual seal force seen with stop-
pered vials or the drop in removal torque of screw-capped
bottles. Heat seals formed between polymeric materials may
become brittle with time and adhesives may lose their bonding
efciency. Other changes in packages are the result of direct
interaction of the product with the package. For instance, a
products solvent system may result in swelling or shrinkage
of components and subsequent improper t. This is the case
when shrinkage of an elastomeric gasket on an aerosol con-
tainer valve results in loss of propellant and/or product.
Package changes that impact integrity may also affect
the packages functionality. For instance, silicone sorption
into a syringes elastomeric plunger may ultimately cause poor
breakaway and extrusion performance. Aged multidose vial
elastomeric closures may lose their ability to effectively reseal
and exhibit increased coring tendencies, both of which are re-
lated to package integrity as well as functionality.
Therefore, package integrity is not complete until the ef-
fects of storage temperature and humidity over shelf life are
monitored. Accelerated aging studies at elevated tempera-
tures can prove valuable in predicting these effects. However,
it is important to verify these results using real-time stability
data. For this reason, primary stability studies used to sup-
port a regulatory submission for new product approval should
include appropriate package integrity and functionality tests.
Marketed Product Shelf Life Stability
The assurance of package integrity should be determined dur-
ing product development by dening the relationships be-
tween leakage rate and package materials/dimensions,
processing/assembly parameters, and product stability/distri-
bution. Once these relationships are established, the institu-
tion of proper quality assurance specications and manufac-
turing controls and monitors can ensure that the components
and processes are kept within allowable limits and will pro-
vide assurance of package integrity. Further leakage tests
should only be necessary to verify control of the processes and
the package. To satisfy regulatory demands, it may be neces-
sary to include appropriate package integrity tests that dem-
onstrate a packages microbial integrity as part of the mar-
keted product stability protocol.
CONCLUSION
In conclusion, package integrity is a simple concept, but one
that is not easily measured or validated. Scientically speak-
ing, leakage is a quantitative term mathematically described
362 Chapter 4
as the amount of gas capable of passing through a seal under
carefully dened conditions of temperature and pressure. In
the same way that nothing is absolutely clean or pure,
most packages exhibit some degree of leakage. In other words,
a leak-tight package is one with leaks so small as to be incon-
sequential. Verifying package integrity is therefore a matter
of dening the leakage specication limits and selecting an
appropriate test method(s) for detecting leakage at these lim-
its. Leakage specications should be conservative enough to
guarantee sterility of the parenteral package and to ensure
satisfactory product stability and package performance
throughout the shelf life, but not so stringent as to make integ-
rity testing verication prohibitively difcult and expensive.
When choosing a leak test method, the sensitivity of the equip-
ment or technique should be weighed against other factors,
such as reliability, reproducibility, speed, cost, user friendli-
ness, data integrity, and data-processing capabilities.
The pharmaceutical industry can no longer take a lacka-
daisical approach to package integrity validation. The growing
demand for novel, more complex package systems has brought
new challenges to package design and development. There are
a variety of package materials, and there are numerous steril-
ization methods available to permit the creation of such
unique package systems. The ability to accurately test pack-
ages for integrity has improved with the variety of package
integrity test methods offered. This is coupled with a greater
understanding of leakage concepts. Finally, new parenteral
product approvals hinge on the data package provided to regu-
latory authorities that satisfactorily demonstrate the pack-
ages ability to function and remain integral over anticipated
conditions of distribution and storage for the life of the
product.
To meet these challenges, leaders in the pharmaceutical
industry are taking a more comprehensive product/package
development approach that incorporates the demonstration
and validation of package integrity and functionality. This be-
gins by careful selection of package component materials and
design. The stability, functional performance, and integrity
expectations of the package need to be dened early in the
development process. This should be followed by the optimiza-
tion of package lling, assembly, and processing operations.
Whenever possible, the integrity and functional performance
of the package should be challenged at the limits of component
dimensional specications and at the operational parameter
extremes of lling, assembly, and processing. Checks are to
be integrated into each critical step to monitor and ensure ade-
quate packaging and process controls. Finished product
should also be evaluated as a function of the distribution and
long-term storage environment. Once product development is
complete, continual monitoring of critical steps in production
serve as the primary tool for assuring package integrity. Addi-
tional leakage tests to demonstrate microbial integrity of the
marketed product over shelf life then provide the nal check
for a thoroughly validated product/package system.
REFERENCES
1. FDA Guidance for Industry, Guidance for the Submission Doc-
umentation for Sterilization Process Validation in Applications
for Human and Veterinary Drug Products (1994).
2. FDA Draft Guidance for Industry, Container and Closure In-
tegrity Testing In Lieu of Sterility Testing as a Component of
the Stability Protocol for Sterile Products (1998).
3. FDA Draft Guidance for Industry, Stability Testing of Drug
Substances and Drug Products (1998).
4. FDADraft Guidance for Industry, INDs for Phase 2 and 3 Stud-
ies of Drugs, Including Specied Therapeutic Biotechnology-
Derived Product, Chemistry, Manufacturing, and Controls
Content and Format (1999).
5. FDA Guidance for Industry, Container Closure Systems for
Packaging Human Drugs and Biologics (1999).
6. Rules Governing Medicinal Products in the European Union,
Pharmaceutical Legislation, Vol. 4, Good Manufacturing Prac-
364 Chapter 4
tices. Annex 1. Manufacture of Sterile Medicinal Products
(1998).
7. M. Carroll et al., Pharmaceutical package integrity, Technical
report 27, PDA J. Pharm Sci. Technol., 52, No. 4, Supplement,
(1998).
8. A. Martin, J. Swarbrick, and A. Cammarata, Physical Phar-
macy: Physical Chemical Principles in the Pharmaceutical Sci-
ences, 3rd ed., Lea & Febiger, Philadelphia, PA, 400, 448453,
532, 537538 (1983).
9. D. K. Morton, Container/closure integrity of parenteral vials,
J. Parenter. Sci. Technol., 41, 145158 (1987).
10. M. A. Amini and D. R. Morrow, Leakage and permeation: the-
ory and practical applications, Package Dev. Syst., 2027
(MayJune 1979).
11. Nondestructive Testing Handbook, Vol. 1, Leak Testing, Ameri-
can Society for Nondestructive Testing, American Society for
Metals, 1982, pp. 612.
Metals, 1982, pp. 57100.
13. L. B. Loeb, The Kinetic Theory of Gases, 3rd ed., Dover Publica-
tions, New York, 1961, pp. 278300.
14. D. Eisenberg and D. M. Crothers, Physical Chemistry and Ap-
plications to the Life Sciences, Benjamin/Cummings Publishing
Co., Menlo Park, CA 1979, p. 726.
15. Introduction to HeliumMass Spectrometer Leak Detection, Var-
ian Assoc., Inc., Palo Alto, CA, 1980.
16. D. K. Morton, N. G. Lordi, and T. J. Ambrosio, Quantitative
and mechanistic measurements of parenteral vial container/
closure integrity. Leakage quantitation, J. Parenter. Sci. Tech-
nol., 43, 8897 (1989).
17. D. K. Morton, N. G. Lordi, L. H. Troutman, and T. J. Ambrosio,
Quantitative and mechanistic measurements of container/clo-
sure integrity. Bubble, liquid, and microbial leakage tests, J.
Parenter. Sci. Technol., 43, 104108 (1989).
18. L. Kirsch, L. Nguyen, and C. Moeckley, Pharmaceutical
container/closure integrity I: mass spectrometry-based helium
leak rate detection for rubber-stoppered glass vials, PDA J.
Pharm Sci. Technol., 51(5), 187194 (1997).
19. L. Kirsch, L. Nguyen, C. Moeckley, and R. Gerth, Pharmaceuti-
cal container/closure integrity II: the relationship between mi-
crobial ingress and helium leak rates in rubber-stoppered glass
vials, PDA J. Pharm Sci. Technol., 51(5), 195201 (1997).
20. H. M. C. Put, H. J. Witvoet, and W. R. Warner, Mechanism
of microbiological leaker spoilage of canned foods: biophysical
aspects, J. Food Prot,, 43, 488497 (1980).
Metals, 1982, p. 620.
22. L. Kirsch, Pharmaceutical container/closure integrity VI: a re-
port on the utility of liquid tracer methods for evaluating the
microbial barrier properties of pharmaceutical packaging, PDA
J. Pharm Sci. Technol., 54(4), 305313 (2000).
23. S. Morris and W. OBrien, Jr., New sensors help improve heat-
seal microleak detection, Packaging Technol. Eng., 4276 (July
1998).
24. F. Moll, D. Doyle, M. Haerer, and D. Guazzo, Validation of a
high voltage leak detector for use with pharmaceutical blow-
ll-seal containersa practical approach, PDA J. Pharm Sci.
Technol., 52(5), 215227 (1998).
25. T. Villiger, Leakage testing of ampuls. Validation method for
high voltage pinhole detectors, Pharm. Inc., 49, 10711074
(1987).
26. D. E. McVean, P. A. Tuerck, G. L. Christenson, and J. T. Car-
stensen, Inadequacies in leakage test procedures for ame-
sealed ampuls, J. Pharm. Sci., 61, 16091611 (1972).
27. R. Jacobus, P. Torralba, J. Moldenhauer, W. Scheidt, and J.
Runkle, Development and validation of a spectrophotometric
dye immersion test method used to measure container-closure
integrity of an oil-based product, PDA J. Pharm Sci. Technol.,
52(3), 110112 (1998).
28. M. S. Korczynski, Evaluation of closure integrity, Aseptic Phar-
maceutical Manufacturing: Technology for the 1990s (W. P.
Olson, Ed.), Interpharm Press, Buffalo Grove, IL (1987).
366 Chapter 4
29. T. E. Odlaug, Microbiological validation of container closure
systems, Pharm. Manuf., 1, 3035 (1984).
30. I. R. Last and K. A. Prebble, Suitability of near infrared meth-
ods for the determination of moisture in a freeze-dried injection
product containing different amounts of the active ingredient,
J. Pharm. Biomed. Anal., 11, 10711076 (1993).
31. G. Birrer, J. Liu, J. Halas, and G. Nucera, Evaluation of a con-
tainer closure integrity test model using visual inspection with
conrmation by near infra-red spectroscopic analysis. PDA J.
Pharm Sci. Technol., 54(5), 373382 (2000).
32. J. Veale, Validation of a tunable diode laser spectrometer for
non-destructive monitoring of oxygen in the headspace of par-
enteral containers, oral presentation at the PDA Annual Meet-
ing, Washington, DC, 2001.
33. V. Wu and B. Crist, An automated method for vacuum testing
pharmaceutical vials, J. Parenter. Sci. Technol., 43, 179182
(1989).
34. B. Crist, Time-dependence of pressure in lyophilized vials, PDA
35. L. Daukas and E. Trappler, Comparative methods of integrity
testing of vials sealed at reduced pressures for lyophiles, PDA
36. D. K. Morton and N. G. Lordi, Residual seal force measurement
of parenteral vials. I. Methodology, J. Parenter. Sci. Technol.,
42, 2329 (1988).
37. J. Ludwig, P. Nolan, and C. Davis, Automated method for de-
termining Instron residual seal force of glass vial/rubber clo-
sure systems, J. Parenteral Sci. Technol., 47(5), 211218
(1993).
38. J. Ludwig and C. Davis, Automated method for determining
Instron residual seal force glass vial/rubber closure systems.
Part II. 13-mm vials, J Parenteral Sci. Technol., 49(5), 253
256 (1995).
39. D. K. Morton and N. G. Lordi, Residual seal force measurement
of parenteral vials, II. Elastomer evaluation, J. Parenter. Sci.
Technol., 42, 5761 (1988).
40. L. Nguyen, W. Muangsiri, R. Schiere, D. Guazzo, and L. Kirsch,
Pharmaceutical container/closure integrity IV: development of
an indirect correlation between vacuum decay leak measure-
ment and microbial ingress, PDA J. Technol., 53(4), 211216
(1999).
41. L. Kirsch, L. Nguyen, A. M. Kirsch, G. Schmitt, M. Koch, T.
Wertli, M. Lehmann, and G. Schramm, Pharmaceutical
container/closure integrity V: an evaluation of the WILCO
LFC method for leak testing pharmaceutical glass-stoppered
vials, PDA J. Pharm Sci. Technol., 53(5), 235239 (1999).
42. 661 Containers, United States Pharmacopeia 25/National
Formulary 20, 19992005 (2002)
43. 671 ContainersPermeation, United States Pharmacopeia
25/National Formulary 20, 20052007 (2002).
44. R. D. Cilento, Methods for evaluating hermetic closures for
screwcapped bottles, J. Pharm. Sci., 66, 333336 (1977).

Parenteral Quality Control

Загружено:

Сведения о документе

Авторское право

Доступные форматы

Поделиться этим документом

Поделиться или встроить документ

Параметры публикации

Этот документ был вам полезен?

Это неприемлемый материал?

Авторское право:

Доступные форматы

Parenteral Quality Control

Загружено:

Авторское право:

Доступные форматы

Marcel Dekker, Inc.

manufactured by the Steris Corporation (formerly

unit used for surface sterilization within

air sampler (courtesy of Bioscience International,

vial/exible container admixture system by Abbott

dual chamber syringe designed to contain both the in

by Vetter GmbH. (Courtesy of Vetter

by West Pharmaceutical Services, Inc.

Luer Injection System by Baxter Healthcare

Вам также может понравиться